KR20160134564A - Device and method for analyzing user emotion - Google Patents

Abstract

The present invention provides a device and a method for analyzing user emotion, capable of analyzing user emotion by analyzing a sentence. According to the present invention, the method comprises the following steps: acquiring a first emotional vocabulary which includes a first keyword group including at least one keyword representing positive emotion and a second keyword group including at least one keyword representing negative emotion; collecting social network data including a third keyword not included in the first emotional vocabulary from a social network; and adding the third keyword to the first or second keyword group based on a frequency that the third keyword and at least one keyword included in the first keyword group co-occur in the social network data, and a frequency that the third keyword and at least one keyword included in the second keyword group co-occur in the social network data, so as to generate a second emotional vocabulary.

Description

TECHNICAL FIELD [0001] The present invention relates to a method and a device for analyzing user's emotions,

And a method and device for analyzing user's emotions.

As social network services became popular, and big data technology developed, there were various attempts to analyze information and obtain social network and chat contents. Especially, as the communication through the online becomes active, it may be necessary to guess the emotion of the user based on the language used by the user. For example, in an automatic dialog system used in a messenger program, a game, or an artificial intelligence UI of a smart phone, an optimal response should be performed by guessing the emotion of a user. Accordingly, an emotional wordbook was created as an attempt to analyze the emotion based on the words inputted by the user, and the word was distributed to the public as open-sourced.

A method and device for analyzing a user's emotions are provided. The present invention also provides a computer-readable recording medium storing a program for causing a computer to execute a method of analyzing a sentence.

A method of a device for analyzing emotions according to an embodiment includes the steps of: generating a first keyword group including at least one keyword representing a positive emotion and a second keyword group including at least one keyword representing a negative emotion; 1) acquiring an emotional wordbook; Collecting social network data including a third keyword not included in the first emotional wordbook from the social network; In the social network data, the frequency with which the third keyword co-occurs with at least one keyword included in the first keyword group and the frequency with which the third keyword coexists with at least one keyword included in the second keyword group, Generating a second emotional wordbook by adding a third keyword to the first keyword group or the second keyword group based on the frequency of the second emoticon wordbook; . ≪ / RTI >

The step of generating the second emotional wordbook may include the step of generating a third emoticon wordbook in the social network data when the frequency with which the third keyword appears in common with at least one keyword included in the first keyword group exceeds a predetermined reference value The third keyword is added to the first keyword group and the third keyword is set to the second keyword group when the frequency in which the third keyword appears in common with at least one keyword included in the second keyword group exceeds a predetermined reference value, To the user.

Further, the first emotional wordbook further includes a third keyword group including at least one keyword indicating neutral emotion, and the step of generating the second emoticon wordbook includes the step of, in the social network data, And adding a third keyword to the third keyword group when the frequency of occurrence in common with at least one keyword included in the third keyword group exceeds a predetermined reference value.

Also, a method for a device to analyze emotions includes receiving chat data from an external device; Determining emotions corresponding to the keywords included in the chat data using the second emotional wordbook; And determining emotions of the user of the external device based on emotions corresponding to the keywords included in the chat data, respectively; As shown in FIG.

In addition, as a method for the device to analyze emotion, The first emotional wordbook and the second emotional wordbook include emotional scores obtained by quantifying the degree of emotions corresponding to at least one keyword included in the first keyword group or the second keyword group,

The step of determining emotion may comprise determining an emotion score for at least one keyword included in the chat data.

In addition, the step of determining emotion may include determining emotion based on a sum of emotion scores for one or more keywords included in the chat data.

The method may further include the step of changing the chatting method with the user based on the determined emotion.

A method for a device to summarize chat content according to an embodiment includes a first group of keywords including at least one keyword associated with a first subject and a second group of keywords including at least one keyword associated with a second subject Obtaining a first specialized wordbook; And a keyword that is not included in the first specialized lexicon in the chat log stored in the device coincides with at least one keyword contained in the second keyword group and a frequency that coincides with at least one keyword included in the first keyword group, Generating a second specialized wordbook by adding a keyword that is not included in the first specialized wordbook to the first keyword group or the second keyword group based on the frequency of occurrence of the second specialized wordbook; . ≪ / RTI >

Also, a method for a device to summarize chat content includes: obtaining chat session data that represents a result of a chat performed with an external device; Extracting one or more core keywords from the keywords included in the chat session data using the second specialized wordbook; And summarizing the contents of the chat session data using one or more core keywords; As shown in FIG.

The step of extracting the core keyword may include extracting a keyword included in the second specialized wordbook among the keywords included in the chat session data.

The step of summarizing the contents may include a step of summarizing the contents into a sentence including the core keyword or summarizing the core keyword and the subject corresponding to the core keyword into a table matching each other.

Further, the step based on the frequency of occurrence of common features is characterized by using pointwise mutual information (PMI).

The device for analyzing emotion according to an embodiment includes a first emotion group including a first keyword group including at least one keyword indicating a positive emotion and a second keyword group including at least one keyword representing a negative emotion, A communication unit for acquiring a wordbook and collecting social network data including a third keyword not included in the first emotional wordbook from the social network; In the social network data, the frequency with which the third keyword appears in common with at least one keyword included in the first keyword group, and the frequency with which the third keyword appears in common with at least one keyword included in the second keyword group A control unit for generating a second emotional wordbook by adding a third keyword to the first keyword group or the second keyword group; A storage unit for storing the first emotional wordbook and the second emotional wordbook; . ≪ / RTI >

In the case where the second emotional wordbook is generated, when the frequency in which the third keyword appears in common with at least one keyword included in the first keyword group exceeds a predetermined reference value, Adding the keyword to the first keyword group; The third keyword may be added to the second keyword group in the case where the frequency in which the third keyword coexists with at least one keyword included in the second keyword group exceeds a predetermined reference value in the social network data.

Further, the first emotional wordbook further includes a third keyword group including at least one keyword indicating neutral emotion, and when the second emotional wordbook is generated, the control unit sets the third keyword in the social network data, The third keyword may be added to the third keyword group when the frequency with which at least one keyword included in the third keyword group exceeds a predetermined reference value.

Further, the communication unit may receive chat data from an external device; The control unit determines the emotion corresponding to each of the keywords included in the chat data using the second emotional wordbook; The emotion of the user of the external device can be determined based on the emotions corresponding to each of the keywords included in the chat data.

The first emotional wordbook and the second emoticonbook word may include a sensitivity score obtained by digitizing the degree of emotion corresponding to at least one keyword included in the first keyword group or the second keyword group, The emotional score for at least one keyword included in the chat data can be determined.

Further, when the emotion is determined, the control unit may determine the emotion based on the sum of emotion scores for one or more keywords included in the chat data.

Further, the control unit can change the chatting method with the user based on the determined emotion.

A device for summarizing chat content according to one embodiment includes a first group of keywords including at least one keyword associated with a first subject and a second group of keywords including at least one keyword associated with a second subject, 1 acquire a specialized vocabulary; In the chat log stored in the device, a keyword that is not included in the first specialized lexicon is common with at least one keyword included in the second keyword group and a frequency appearing in common with at least one keyword included in the first keyword group A control unit for generating a second specialized wordbook by adding keywords not included in the first specialized wordbook to the first keyword group or the second keyword group based on the frequency; And a storage unit for storing the first specialized wordbook and the second specialized wordbook; . ≪ / RTI >

In addition, the control unit may acquire chat session data indicating a result of the chat performed with the external device; Extract one or more core keywords from the keywords included in the chat session data using the second specialized wordbook; The content of the chat session data can be summarized using one or more key keywords.

In addition, when the core keyword is extracted, the control unit may extract a keyword included in the second specialized wordbook among the keywords included in the chat session data.

The control unit may summarize the contents into a sentence including the core keyword, or summarize the keywords and the topics corresponding to the core keywords in a table that matches each other. In addition, The PMI is used.

There is provided a computer-readable recording medium storing a program for causing a computer to execute a sensitivity analysis method according to an exemplary embodiment.

1 is a block diagram illustrating a system for analyzing emotions of a user according to an embodiment.
2A is a diagram illustrating an emotional wordbook according to an exemplary embodiment of the present invention.
FIG. 2B is a diagram showing an example of an emotional word list including emotional scores.
FIG. 2C is a diagram illustrating a specialized linguistic dictionary according to an embodiment.
3 is a block diagram briefly illustrating the structure of a device according to one embodiment.
4 is a block diagram illustrating a device according to one embodiment.
5 is a flowchart illustrating a method of analyzing emotions of a user of an external device according to an exemplary embodiment of the present invention.
6 is a flowchart illustrating a method for a device according to an embodiment to generate an emotional wordbook.
FIG. 7 is a flow diagram illustrating a method for a device according to an embodiment to create a specialized linguistic dictionary.
8 is a diagram illustrating an example of displaying social network data according to an embodiment.
FIG. 9 is a flowchart showing more specifically step 640 of the method for generating the emotional wordbook shown in FIG.
10 is a flow chart showing more specifically step 740 of the method for generating the specialized lingua franca shown in Fig.
11 is a flowchart illustrating a method for a device according to an embodiment to analyze the emotion of an external device user.
12 is a flow diagram illustrating a method for a device according to an embodiment to summarize a chat session.
13 is a diagram illustrating a chat session in accordance with one embodiment.
14 is a flow chart showing more specifically step 1110 of FIG.
15 is a diagram specifically showing steps 1120 and 1130 of FIG.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention. However, the embodiments may be embodied in various different forms and are not limited to the embodiments described herein. In the drawings, the same reference numbers are used throughout the specification to refer to the same or like parts.

Throughout the specification, when a part is referred to as being "connected" to another part, it includes not only "directly connected" but also "electrically connected" with another part in between . It is also to be understood that when an element is referred to as "comprising ", it does not exclude other elements unless specifically stated to the contrary.

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

1 is a block diagram illustrating a system for analyzing emotions of a user according to an embodiment.

In one embodiment, the system may include a server 100, a user device 120, and an external device 140. In another embodiment, the system may include only user device 120 and external device 140. The user device 120 may communicate with the external device 140. Communication between the user device 120 and the external device 140 may be performed using the communication network of the server 100 or the communication company or may be performed using the direct connection network 103. [ Communication between the user device 120 and the external device may be performed using local communication such as Bluetooth or infrared rays. When a user device 120 and an external device 140 are connected through a direct connection network 103, a program necessary for performing communication with at least one of the user device 120 and the external device 140 may be installed have. In this case, the result of performing the communication may be stored in at least one of the user device 120 and the external device 140. When the user device 120 and the external device 140 communicate with each other through the server 100, it is necessary to perform communication with at least one of the server 100, the user device 120 and the external device 140 The program may be installed. In this case, the result of performing the communication may be stored in at least one of the server 100, the user device 120, and the external device 140. For example, a program for performing a chat may be installed in at least one of the server 100, the user device 120, and the external device 140. In this case, a chat log indicating the result of chatting may be stored in at least one of the server 100, the user device 120, and the external device 140.

In one embodiment, the user device 120 may be utilized to provide customer service. For example, a user of the user device 120 may be a customer agent and may provide a customer consultation service via chat using the user device 120. [

In one embodiment, the server 100 or the user device 120 may analyze the emotion of the user of the external device 140. In this case, the server 100 or the user device 120 can acquire the emotional wordbook necessary for analyzing the emotion of the user of the external device 140. The server 100 or the user device 120 can acquire the existing first emotional wordbook and collect social network data including new keywords not included in the first emotional wordbook from the social network. The server 100 or the user device 120 may add a new keyword included in the social network data to the first emotional wordbook to generate a second emotional wordbook that extends the first emotive wordbook. The server 100 or the user device 120 can analyze the emotion of the user of the external device 140 using the second emotional wordbook. In one embodiment, server 100 or user device 120 may summarize the content of a chat performed by user device 120 and external device 140. In this case, the server 100 or the user device 120 may acquire a specialized linguistic linguistic dictionary and summarize the content of the chat performed by the user device 120 and the external device 140 using the acquired linguistic linguistic dictionary . If the existing specialized wordbook does not exist, the server 100 or the user device 120 may clustering the chat log to generate the first specialized wordbook. In the case where the existing specialized wordbook exists, the server 100 or the user device 120 can acquire the existing first specialized wordbook. The server 100 or the user device 120 may obtain new keywords not included in the first specialized lexicon from the chat log. The server 100 or the user device 120 may add a new keyword to the first specialized lexicon to generate a second specialized linguistic dictionary extending the first specialized linguistic lexicon. The server 100 or the user device 120 may summarize the content of the chat performed by the user device 120 and the external device 140 using the second specialized lexicon.

2A is a diagram illustrating an emotional wordbook according to an exemplary embodiment of the present invention. The emotional wordbook 200 may include an affirmative table 201 and a negative table 202. The emotional wordbook 200 may further include a neutral table (not shown). The affirmation table 201 may include one or more keywords (e.g., Happy, Great, Good, etc.) indicating a positive emotion and the negative table 202 may include one or more keywords (e.g., Sad, Bad, Terrible, etc.). The neutral table may include one or more keywords that represent neutral emotions, i. E. Do not exhibit positive emotions and negative emotions. Initially, the server 100 or the user device 120 may obtain an open-source emotional wordbook. Open Source means information that is freely available to anyone. Server 100 or user device 120 may extend emotional wordbook 200 using the methods described herein. The open source emotional vocabulary is not only publicly available for free, but can also be updated by the public. Accordingly, the server 100 or the user device 120 can directly acquire the updated open source emotional wordbook in addition to expanding the emotional wordbook 200 using the method disclosed herein, It can also be extended.

FIG. 2B is a diagram showing an example of an emotional word list including emotional scores. The emotional wordbook 220 may include a keyword table 221 and a corresponding emotional score table 222. In one embodiment, the score corresponding to the keyword indicating the positive emotion may be a positive number, and the score corresponding to the keyword indicating the negative emotion may be negative. The score corresponding to the keyword may have a larger absolute value as the keyword exhibits a stronger emotion. The score corresponding to the keyword may have a smaller absolute value as the keyword indicates a feeling of being close to neutral. In one embodiment, the keywords included in the emotional wordbook 220 may correspond to an emotional score between -1 and 1 inclusive. Initially, the server 100 or the user device 120 may obtain an open source emotional wordbook free of charge in the public domain. The server 100 or the user device 120 may extend the emotional wordbook 220 using the methods disclosed herein. The open source emotional vocabulary is not only publicly available for free, but can also be updated by the public. Thus, in addition to extending the emotional wordbook 220 directly using the method described herein, the server 100 or the user device 140 may obtain the updated open-source emotional wordbook to update the emotional wordbook 220 It can also be extended.

In one embodiment, the server 100 or user device 120 may set the sensitivity score differently depending on the application. For example, the emotional wordbook 220 may be used for automatic chatting for customer center automation. Since the chat contents in the customer center are mainly related to inquiries about product failure or error, many keywords indicating negative emotions may be used. Accordingly, the server 100 or the user device 120 can relax the criterion for determining the negative emotional state of the emotional wordbook 220.

For example, in the open source emotional wordbook, the emotional score of the keyword "Bad" may be -0.6 indicating negative emotions. However, in a chat in a customer center, keywords such as "Bad" can be frequently used to indicate a product failure or error regardless of the user's feelings. Therefore, "Bad" can be set to 0 point in the emotional wordbook used in the customer center.

FIG. 2C is a diagram illustrating a specialized linguistic dictionary according to an embodiment. The specialized lexicon 260 may include a group of keywords related to one or more common subjects. As shown in FIG. 2C, the specialized lexicon 260 includes a "model" table 261 that contains at least one keyword (eg, A, B, C, etc.) associated with the "model" Includes at least one keyword (reset, vaccine, genuine use, etc.) associated with the "problem" table 262 and at least one keyword associated with the "solution", including at least one keyword (e.g., update, Solution "table < RTI ID = 0.0 > 263 < / RTI > Initially, the server 100 or the user device 120 may obtain a chat log stored in the server 100 to create a first specialized lexicon. The server 100 or the user device 120 may clustering chat logs to generate keyword groups related to similar topics. Clustering is a technique of data mining, in which information with similarity in a database can be grouped into several groups called clusters. The server 100 or the user device 120 may generate a table using clusters related to a similar topic to generate an initial specialized lingua lang.

3 is a block diagram briefly illustrating the structure of a device according to one embodiment. The device 300 may correspond to the user device 120 or the server 100 of FIG. The device 300 may include a communication unit 302, a control unit 304, and a storage unit 306.

The communication unit 302 communicates with the external device 140 under the control of the control unit 304 and can transmit or receive information. The communication unit 302 may perform communication using one of a 2G network, a 3G network, an LTE network, Wifi, Bluetooth, infrared, and ZigBee, but is not limited thereto.

The communication unit 302 can collect information using the Internet network. For example, the communication unit 302 may collect social network data. In one embodiment, the communication unit 302 can acquire a keyword as a keyword indicating a positive emotion and an emotion wordbook that classifies the keyword as a keyword indicating negative emotion. The emotional wordbook may further include a keyword indicating neutral emotions, and may include emotional scores obtained by quantifying the degree of emotions corresponding to the keywords included in the emotional wordbook. Emotional wordbooks can be open source emotional wordbooks that are free to the public. An example of an emotional wordbook according to an embodiment is shown in Figs. 2A and 2B.

In another embodiment, the communication unit 302 may acquire a chat log indicating a result of a plurality of devices performing chat for a predetermined period or more. In another embodiment, the communication unit 302 can acquire a specialized wordbook. The specialized wordbook may be a wordbook containing a group of keywords related to one or more common topics.

The storage unit 306 may store at least one of an emotional wordbook, a specialized wordbook, and a chat log under the control of the control unit 304. [ The specialized wordbook may be one received by the communication unit 302 or generated by the control unit 304. [ The chat log may be a result of directly performing the chat by the communication unit 302 or may be the result of the communication unit 302 receiving the result of the chat by one or more other devices. The storage unit 306 may store control programs and applications for controlling the device 300 or the control unit 304. [

The control unit 304 can control the communication unit 302 and the storage unit 306. The control unit 304 analyzes and processes the information received by the communication unit 302 and controls the storage unit 306 to store the received information. In one embodiment, the control unit 304 may add a new keyword included in the social network data to the first emotional wordbook, thereby generating a second emotional wordbook that extends the first emotive wordbook. The control unit 304 can analyze the emotion of the user of the external device 140 using the second emotional wordbook. In another embodiment, the control 304 may obtain new keywords not included in the first specialized lexicon from the chat log. The server 100 or the user device 120 may add a new keyword to the first specialized lexicon to generate a second specialized linguistic dictionary extending the first specialized linguistic lexicon. The server 100 or the user device 120 may summarize the content of the chat performed by the user device 120 and the external device 140 using the second specialized lexicon.

4 is a block diagram illustrating a device according to one embodiment. 4 is a device according to one embodiment, and the device may include fewer components or may include more components than the components shown in FIG.

4, the device 300 includes a control unit 304, a mobile communication module 320, a sub communication module 330, an input / output module 360, a sensor module 370, a storage unit 306, And a display unit 390. The sub communication module 330 may include at least one of a wireless LAN module 331 and a local communication module 332. The input / output module 360 may include at least one of a button 361, a microphone 362, a speaker 363, a vibration motor 364, and a touch panel 365. The communication unit 302 of FIG. 3 may include at least one of a mobile communication module 320 and a sub communication module 330.

The control unit 304 stores the ROM 312 storing the control program for controlling the device 300 and the signal or data input from the outside of the device 300, A RAM 313 used as a storage area for the operation and a GPU 314 for processing graphics of the device 300. [ The CPU 311 may include a plurality of processors such as a single core, a dual core, a triple core, or a quad core. The CPU 311, the ROM 312, the RAM 313, and the GPU 314 may be interconnected via an internal bus (BUS).

The control unit 304 may control the mobile communication module 320, the sub communication module 330, the input / output module 360, the sensor module 370, the storage unit 306, and the display unit 390.

The mobile communication module 320 may allow the device 300 to be connected to an external device through mobile communication using at least one (one or more) antennas (not shown) under the control of the controller 304. The mobile communication module 320 is connected to a mobile phone (not shown), a smart phone (not shown), a tablet PC (not shown) or other device (not shown) And can transmit / receive a radio signal for making a call, a video call, a text message (SMS) or a multimedia message (MMS) transmission.

The sub communication module 330 may include at least one of a wireless LAN module 331 and a local area communication module 332. For example, it may include only a wireless LAN module 331, only a short range communication module 332, or both a wireless LAN module 331 and a short range communication module 332.

The wireless LAN module 331 can connect to the Internet at a place where a wireless access point (not shown) is installed under the control of the controller 304. [ The wireless LAN module 331 can support IEEE 802.11x of the IEEE Institute of Electrical and Electronics Engineers (IEEE). The short-range communication module 332 can perform short-range wireless communication between the device 300 and the image forming apparatus (not shown) under the control of the control unit 304. [ The short-distance communication method may include Bluetooth, infrared data association (IrDA), Zig-bee method, and the like.

The device 300 may include at least one of a mobile communication module 320, a wireless LAN module 331, and a local communication module 332 depending on its performance.

The input / output module 360 may include at least one of a plurality of buttons 361, a microphone 362, a speaker 363, a vibration motor 364, and a touch panel 365.

Button 361 may be formed on the front, side, or rear of the housing of device 300 and may include a power / lock button (not shown), a volume button (not shown), a home button, a back button, and a search button 361.

The microphone 362 may receive voice or sound according to the control of the controller 304 to generate an electrical signal.

The speaker 363 can output a sound corresponding to various signals of the mobile communication module 320 and the sub communication module 330 to the outside of the device 300 under the control of the control unit 304. [ The speaker 363 can output a sound corresponding to the function performed by the device 300. [ The speaker 363 may be formed in one or more of the appropriate positions or positions of the housing of the device 300.

The vibration motor 364 can convert the electrical signal into mechanical vibration under the control of the control unit 304. [ For example, the device 300 in the vibration mode may operate the vibration motor 364 when a voice call is received from another device (not shown). The vibration motor 364 may operate in response to the touching operation of the user touching the touch panel 365 and the continuous movement of the touch input on the touch panel 365.

The touch panel 365 may transmit an analog signal corresponding to at least one touch input to the user interface to the control unit 304. The touch panel 365 may receive at least one touch input via a user's body (e.g., a finger) or touchable input means (e.g., a stylus pen). In addition, the touch panel 365 may receive continuous movement of at least one touch input. The touch panel 365 can transmit an analog signal corresponding to the continuous movement of the touch input to the control unit 304. [

The touch input is not limited to input through contact between the touch panel 365 and the user's body or touchable input means, but may be a non-contact input (e.g., a distance between the touch panel 365 and the body is 1 mm or less) . ≪ / RTI > The interval that can be detected by the touch panel 365 can be changed according to the performance or structure of the device 300. [

The touch panel 365 may be implemented using, for example, a resistive method, a capacitive method, an infrared method, or an ultrasound wave method. The touch panel 365 can transmit the result of converting the received analog signal to a digital signal (for example, an X coordinate and a Y coordinate) to the controller 304. The control unit 304 can control the touch panel 365 using the digital signal received from the touch panel 365. For example, the control unit 304 may select an application execution icon (not shown) displayed on the touch panel 365 or execute an application in response to the touch input.

The sensor module 370 includes at least one sensor for detecting the state of the device 300. [ For example, the sensor module 370 may include a proximity sensor that detects whether a user accesses the device 300, a light sensor (not shown) or device 300 that detects the amount of light around the device 300, (Not shown) that detects motion of the device 300 (e.g., rotation of the device 300, acceleration or vibration applied to the device 300). The sensor of the sensor module 370 may be added or deleted depending on the performance of the device 300. [

The storage unit 306 controls the operations of the mobile communication module 320, the sub communication module 330, the input / output module 360, the sensor module 370, and the display unit 390 under the control of the control unit 304 And can store signals or data correspondingly input / output. The storage unit 306 may store control programs and applications for controlling the device 300 or the control unit 304. [

The term "storage unit" may include a storage unit 306, a ROM 312 in the control unit 304, a RAM 313, or a memory card (not shown) mounted in the device 300. The storage unit may include a nonvolatile memory, a volatile memory, a hard disk drive (HDD), or a solid state drive (SSD).

The display unit 390 displays and outputs information processed by the device 300 on the display panel 391. For example, the display unit 390 may display a user interface for selecting a virtual image, a user interface for setting an operation of a virtual image, and a user interface for purchasing an item of a virtual image.

Meanwhile, when the display panel 391 and the touch panel 365 have a layer structure and are configured as a touch screen, the display unit 390 can be used as an input device in addition to the output device. The display panel 391 of the display unit 390 may be a liquid crystal display, a thin film transistor-liquid crystal display, an organic light-emitting diode, a flexible display ), A three-dimensional display (3D display), and an electrophoretic display (electrophoretic display). The device 300 may include two or more display units 390 depending on the implementation of the device 300. [ In this case, the two or more display units 390 may be arranged to face each other using a hinge.

5 is a flowchart illustrating a method of analyzing emotions of a user of an external device according to an exemplary embodiment of the present invention. In one embodiment, the user device 120 and the external device 140 may communicate via the server 100. [ In one embodiment, the user of the external device 140 may be a customer, and the user of the user device 120 may be the entity providing the customer service. The subject providing the customer service may be a person or a device equipped with a program capable of providing an automated service. The server 100 can determine the emotion of the user of the external device 140 or summarize the contents of the chat performed by the user device 120 and the external device 140. [

In step 500, the server 100 may obtain an emotional word list or a specialized word list. The server 100 can directly generate an emotional wordbook or a specialized wordbook or receive a wordbook already generated from the outside.

In steps 510 and 520, the user device 120 and the external device 140 may initiate a chat. The chat may be initiated by the external device 140 requesting a chat with the user device 120.

In steps 512 and 522, the user device 120 and the external device 140 may perform a chat. The chat may be performed using a chat program installed in at least one of the user device 120, the external device 140, and the server 100. [ In step 502, the server 100 may determine the emotion of the user of the external device 140 by analyzing the results of the user device 120 and the external device 140 performing the chat in steps 512 and 522. [ The server 100 may transmit the determined emotion to the user device 120 at step 504. [

In one embodiment, the user device 120 may be a device on which a program capable of providing automated services is installed. In this case, the user device 120 can perform an automatic chat with the external device 140. The user device 120 can receive the emotion of the user of the external device 140 from the server 100 while performing an automatic chat.

In one embodiment, the user device 120 may change the chatting method based on the emotion of the user of the external device 140. For example, the negative emotion of the user of the external device 140 may exceed a negative threshold. In this case, the user device 120 may terminate the automatic chat and allow the user of the user device 120 to conduct a chat with the user of the external device 140 directly.

In one embodiment, the server 100 may perform an automatic chat with a user of the external device 140. The server 100 can determine the emotion of the user of the external device 140 while performing an automatic chat. In one embodiment, the server 100 may change the chatting method based on the emotion of the user of the external device 140. For example, the negative emotion of the user of the external device 140 may exceed a negative threshold. In this case, the server 100 may terminate the automatic chat and call the user device 120. [ The server 100 may allow a user of the user device 120 to directly conduct a chat with a user of the external device 140. [

In step 502, the server 100 also analyzes the results of the user device 120 and the external device 140 performing the chat in steps 512 and 522, so that the user device 120 and the external device 140 chat You can summarize what you have done. The server 100 may also send the summarized content to the user device 120 at step 504.

In steps 514 and 524, the user device 120 and the external device 140 may terminate the chat. The chat may be terminated when one of the user device 120 and the external device 140 receives a chat end command.

6 is a flowchart illustrating a method for a device according to an embodiment to generate an emotional wordbook.

The device 300 according to one embodiment may correspond to the server 100 or the user device 120. [ As shown in FIG. 6, in step 610, the device 300 may obtain a first emotional wordbook. At this time, the first emotional wordbook may be an open source emotional wordbook, and open source means information that is open to the public free of charge. The first emotional wordbook may include a keyword group representing positive emotions and a keyword group representing negative emotions. In addition, the emotional wordbook may further include a keyword group indicating neutral emotions. The emotional wordbook may include a sensitivity score obtained by quantifying the degree of emotional corresponding to each keyword. For example, the emotional score may range from a maximum of 1 to a minimum of -1. In this case, the emotion score of the keyword showing the most positive emotion can be one point, and the emotion score of the keyword showing the most negative emotion can be -1 point. The sensibility score of the keyword indicating the neutral sensibility may be zero. An example of an emotional wordbook is shown in Figures 2a and 2b.

In step 620, the device 300 may collect social network data. The social network data may refer to data stored in a social network service server. For example, the social network data may include text, image, video, and audio data updated by users of the social network service server using the social network service. In this specification, the social network data refers to text data among data stored in a social network server.

The device 300 may perform crawling on one or more social network service web pages to collect social network data. Crawling is a technique used in data mining, which refers to a technique of collecting data stored in a distributed manner on a plurality of computers or servers. An example for a social network web page is shown in FIG.

In operation 630, the device 300 may divide the social network data collected in operation 620 into keywords. The device 300 may use an n-gram method for separating social network data by keyword. For example, in the 1-gram method, the device 300 can separate a sentence included in the social network data and generate a plurality of keywords each consisting of a word. In another example, the device 300 may separate a sentence contained in the social network data using a 2-gram scheme to generate a plurality of keywords consisting of two words.

In step 640, the device 300 may calculate a co-occurrence frequency of at least one keyword included in the new keyword and the emotional wordbook created in step 630, and add a new keyword to the emotive wordbook. For example, the frequency at which the new keyword generated in step 630 coincides with the keyword indicating the positive emotion contained in the emotional wordbook in the social network data may exceed a predetermined reference value. In this case, the device 300 can add a new keyword to the emotional wordbook as a keyword indicating positive emotion. The common appearance of a plurality of keywords may mean that a plurality of keywords are commonly included in the same corpus in the social network data. The corpus may refer to a set of keywords that underlies the analysis of the big data. For example, each of the posts 810-840 posted on the social network service web page, shown in FIG. 8, Lt; / RTI > Accordingly, when a plurality of keywords are commonly included in each of the posts 810 to 840, it can be determined that a plurality of keywords appear in common.

 For example, "good" may be one of the keywords expressing a positive emotion contained in the emotion wordbook. And "pleasant" may be a keyword not included in the first emotional word list among the keywords included in the social network data. 11, the device 300 can determine that "good" and "enjoyable" coexist when common "good" and "pleasant" are included in one post 1110. In one embodiment, the frequency of common appearances with the keywords indicating the positive emotional state of the "pleasant" emotional wordbook may exceed a predetermined reference value. In this case, the device 300 can add "pleasant" as a keyword indicating a positive emotion of the emotional wordbook.

In one embodiment, the device 300 may obtain a sensitivity score that quantifies the degree of emotion corresponding to "pleasant ". The emotional score corresponding to "pleasant" can be determined according to the frequency with which the words "pleasant" appear in common with the keywords indicating the positive emotions of the emotional wordbook. In another embodiment, the device 300 may perform the same method of analysis for keywords indicating negative emotions or for keywords representing neutral emotions.

As a result of performing step 640, the device 300 can acquire an extended emotional wordbook (second emotional wordbook) including the same or more keywords than the first emotive wordbook. The device 300 may periodically or continuously add keywords to the emotional wordbook using the method shown in FIG.

FIG. 7 is a flow diagram illustrating a method for a device according to an embodiment to create a specialized linguistic dictionary. The device 300 may correspond to the server 100 or the user device 120. In step 710, if there is an existing specialized linguistic dictionary generated, the device 300 can acquire the existing first specialized linguistic linguistic dictionary. If there is no existing specialized wordbook that has been created, the device 300 may generate the first first specialized wordbook. The specialized wordbook may be a wordbook containing a group of keywords related to one or more common topics. An example of a specialized wordbook is shown in Figure 2c.

At step 720, the device 300 may collect the chat log. The chat log may be data indicating a result of a plurality of devices including the user device 120 and the external device 140 performing a chat through the server 100. [ If there are no existing specialized wordbooks, the device 300 may collect all the chat logs stored in the server 100. If there is an existing specialized wordbook, the device 300 can acquire a newly generated chat log only after the time at which the device 300 collects the chat log to expand or create the latest specialized wordbook.

In step 730, the device 300 may separate the chat log collected in step 720 on a keyword-by-keyword basis. The device 300 can use the n-gram method to separate the chat log by keyword unit. For example, in the 1-gram method, the device 300 can separate a sentence included in a chat log to generate a plurality of keywords composed of one word. In another example, the device 300 may generate a plurality of keywords consisting of two words by separating the sentences contained in the chat log using the 2-gram method.

In step 740, the device 300 may calculate a common occurrence frequency of the new keyword generated in step 730 and at least one keyword included in the first specialized wordbook, and add a new keyword to the first specialized wordbook. For example, the frequency at which the new keyword generated in step 730 coincides with the keyword related to the specific topic included in the specialized lyric word in the chat log may exceed a predetermined reference value. In this case, the device 300 may add a new keyword to the specialized lexicon as a keyword related to a specific topic. The common appearance of a plurality of keywords may mean that a plurality of keywords are commonly included in the same corpus in the chat log. For example, it may mean that a plurality of keywords are commonly included in one chat session. One chat session may include chat content from the time the chat between the user device 120 and the external device 140 starts until the chat session ends. As a result of performing step 740, the device 300 may obtain an extended specialized lexicon (second specialized lexicon) containing the same or more keywords than the first specialized lexicon.

8 is a diagram illustrating an example of displaying social network data according to an embodiment. The social network data may be separately displayed in units of posts 810 to 840. [ The posts 810 to 840 may be written by the same user or by a plurality of users. The social network data may include keywords contained in the posts 810 through 840. The posts 810 to 840 can be used as a criterion for calculating the common appearance frequency between the keyword of the social network data and the keyword of the emotional wordbook. For example, keywords in the social network data and keywords indicating the positive emotional state of the emotional wordbook may appear in the same posts 810 to 840 in common. In this case, it can be determined that the keyword of the social network data and the keyword showing the positive emotion of the emotional wordbook appear in common.

The device 300 may consider the irony when adding the keyword of the social network data to the emotional wordbook. For example, keywords indicating "unkindness", "anger" and "expensive" are included as keywords indicating negative emotions of the emotional wordbook, and "pleasant" may be included as a keyword showing positive emotions. In this case, the device 300 may analyze the social network data to obtain a new keyword "good " that is not included in the existing emotional wordbook. The keyword "good" appears in common with the keyword "joyous" in the post 810, and appears in common with the word " unkind, " In the post 840 case, "good" may be the keyword chosen to use the irony if "good" in other articles has a high frequency of appearing in common with other keywords representing positive emotions. In this case, the device 300 can generate a more accurate emotional wordbook by excluding the common emotional record in the post 840. For example, the device 300 may exclude the occurrence frequency of a keyword used in the irony method from the total appearance frequency.

In one embodiment, the device 300 may add a keyword indicative of a positive emotion associated with a particular keyword or a keyword indicative of a negative emotion to the emotional wordbook to generate an emotional wordbook specific to the particular keyword. For example, the device 300 may need to find a keyword indicating a positive emotion associated with "Mobile Phone A" or a keyword indicating a negative emotion. In this case, the device 300 can calculate the common appearance frequency between the keyword "Mobile phone A" and the keyword included in the social network data. Further, the device 300 can calculate the common appearance frequency between the keyword "Mobile phone A" and the keyword included in the emotional wordbook. For example, the "fast" keyword may indicate positive emotion, neutral emotion or negative emotion depending on the context. However, when the keyword "fast" and the keyword "mobile phone A" appear in common, the keyword "fast" may mean that the performance of the mobile phone is fast. Therefore, the device 300 can determine the keyword "fast " as a keyword indicating a positive emotion. In one embodiment, the device 300 may obtain a new keyword " fast "from the social network data that is not included in the emotional word list. In this case, the device 300 can analyze the frequency with which the keywords "fast" and "cell phone A" appear in common with the keywords included in the existing emotional wordbook. For example, the frequency of common occurrence with a plurality of keywords indicating positive emotions included in the existing emotional wordbook may exceed a predetermined reference value, for example, the keywords "mobile phone A" and "fast". In this case, when the "fast" keyword is used with the "mobile phone A" keyword, the device 300 can determine that the keyword "fast" is a keyword indicating positive sentiment.

FIG. 9 is a flowchart showing more specifically step 640 of the method for generating the emotional wordbook shown in FIG. At step 900, the device 300 may obtain the keyword separated in step 630 of FIG.

In step 910, the device 300 may calculate the common occurrence frequency of at least one keyword included in the keyword acquired in step 900 and the emotional wordbook. In one embodiment, pointwise mutual information (PMI) may be used to calculate the common occurrence frequency. Let p (x) be the probability of keyword x appearing in the corpus, p (y) be the probability of appearance of y, and p (x, y) The PMI value can be calculated using the following equation.

The method of calculating the common appearance frequency of a keyword using the PMI is the same as that of the conventional technique.

In step 920, the device 300 may extract a keyword having the common occurrence frequency calculated in step 910 equal to or higher than the reference value. For example, the new keyword " awesome " may have a common frequency of occurrence of the keywords "happy "," great ", and "good " In this case, the device 300 may add "awesome " as a keyword indicating a new positive emotion of the emotional wordbook. Likewise, the new keyword " painful "may have a common frequency of occurrence of the keywords" sad ", "bad "," terrible ", etc. indicating negative emotions included in the existing emotional wordbook, In this case, the device 300 may add "painful" as a keyword indicating a new negative emotion of the emotional wordbook.

In step 950, the device 300 may extend the existing emotional wordbook by adding the keyword extracted in step 920 to the existing emotive wordbook. Device 300 may perform step 950 and then go back to step 900 to analyze other keywords.

10 is a flow chart showing more specifically step 740 of the method for generating the specialized lingua franca shown in Fig. In step 1000, the device 300 may obtain the keyword separated in step 730 of FIG.

In operation 1010, the device 300 may calculate a common appearance frequency of at least one keyword included in the keyword and the sensory wordbook obtained in operation 1000. In one embodiment, PMI can be used to calculate the common occurrence frequency. Let p (x) be the probability of keyword x appearing in the corpus, p (y) be the probability of appearance of y, and p (x, y) The PMI value can be calculated using the following equation.

The method of calculating the common appearance frequency of a keyword using the PMI is the same as that of the conventional technique.

In step 1020, the device 300 may extract a keyword whose common occurrence frequency calculated in step 1010 is equal to or higher than a reference value.

For example, the new keyword " D " may have a common frequency of occurrence, such as keywords "A "," B ", and "C " associated with the" model "subject included in the existing specialized lexicon, In this case, the device 300 may add "D" as a keyword associated with the new "model" Likewise, the new keyword "WiFi" may have a common frequency of occurrence, such as the keywords "power "," screen ", "battery ", etc., associated with the" problem " In this case, the device 300 may add "Wi-Fi" as a keyword associated with the new "problem"

In step 1050, the device 300 may extend the existing emotional wordbook by adding the keyword extracted in step 1020 to the existing emotive wordbook. The device 300 may perform step 1050 and then go back to step 1000 to analyze other keywords.

11 is a flowchart illustrating a method for a device according to an embodiment to analyze the emotion of an external device user. The device 300 according to one embodiment may correspond to the server 100 or the user device 120. [

Referring to FIG. 11, in step 1110, the device 300 may acquire chat data indicating a result of performing a chat between the user device 120 and the external device 140, and may separate the chat data in units of keywords . The method of separating chat data by keyword may be performed using an n-gram method.

In step 1120, the device 300 may determine the emotion corresponding to each of the keywords separated in step 1110. For example, the keyword extracted in step 1110 may be included as a keyword indicating a positive emotion in the emotion wordbook (second emotion wordbook) generated in FIG. In this case, the device 300 can determine that the keyword separated in step 1110 is a keyword indicating a positive emotion. In another embodiment, the device 300 may use the emotional wordbook (second emotional wordbook) to determine the emotional score corresponding to the keyword separated in step 1110.

The method of generating the emotional wordbook of Fig. 6 and the method of analyzing emotions of Fig. 11 may be performed in the same device or may be performed in different devices. For example, the server 100 may generate an emotional wordbook, and the user device 120 may analyze emotions. In another example, the server 100 or the user device 120 may generate emotional wordbooks and analyze emotions.

In step 1130, the device 300 may determine the emotion of the user of the external device 140 based on the emotion corresponding to each keyword, which was determined in step 1120. [ For example, the device 300 may determine that the keyword indicating the positive emotion among the keywords determined in the step 1120 is greater than the keyword indicating the negative emotion. In this case, the device 300 can determine that the user of the external device 140 is showing a positive emotion. In another example, the device 300 may determine that the sum of the emotion scores of each of the keywords determined in step 1120 is positive. In this case, the device 300 can determine that the user of the external device 140 is showing a positive emotion. In addition, the device 300 can determine that the greater the absolute value of the sum of the sensibility scores, the stronger the user of the device 140 is. The device 300 may determine that the sum of the emotion scores of each of the keywords determined in step 1120 is negative. In this case, the device 300 can determine that the user of the external device 140 indicates a negative emotion. In addition, the device 300 can determine that the greater the absolute value of the sum of the sensibility scores, the stronger the user of the device 140 is.

In one embodiment, the device 300 may analyze the emotions of the user of the external device 140 after the user device 120 and the external device 140 terminate the chat. In another embodiment, the device 300 may analyze the user's emotions while the user device 120 and the external device 140 are performing a chat. For example, each time the device 300 receives the chat data from the external device 140, the device 300 may analyze the emotion of the user of the external device 140 by separating the chat data by keyword unit. In another example, each time the device 300 receives chat data from the external device 140 for a predetermined number of times or more, the device 300 may analyze the emotion of the user of the external device 140 by separating the chat data by keyword unit. In another embodiment, the device 300 may analyze the emotion of the user of the external device 140 by separating the chat data received from the external device 140 for a predetermined time period by keyword unit.

12 is a flow diagram illustrating a method for a device according to an embodiment to summarize a chat session. The device 300 may correspond to the server 100 or the user device 120.

In operation 1210, the device 300 may obtain the chat session data indicating the result of performing the chat by the user device 120 and the external device 140, and may then separate the chat session data by keyword. The chat session data may refer to all chat data exchanged between when the first user and the second user initiate a chat, and when they end. The method of separating the chat session data by keyword may be performed using an n-gram method. An example of how the user device 120 and the external device 140 perform the chat is shown in FIG.

In step 1220, the device 300 may analyze the keyword separated in step 1210 to select a core keyword. The core keyword may be a keyword corresponding to each of the topics included in the specialized wordbook. The device 300 can extract a core keyword corresponding to each of the topics included in the specialized linguistic word among the keywords separated in step 1210. [ For example, referring to FIG. 16, the device extracts keywords that are the same as or similar to "update", "screen", or "battery", which are keywords corresponding to the "problem" Can be detected. The device 300 may determine that there is a keyword corresponding to the "problem" subject among the keywords separated in step 1210. [ In this case, the device 300 can extract the keywords corresponding to the "problem" topic among the keywords separated in step 1210 as key keywords corresponding to the "problem"

The method for generating the specialized lyric words in Fig. 7 and the method for summarizing the chat session in Fig. 12 may be implemented in the same device, or may be implemented in different devices. For example, the server 100 may generate a specialized lexicon and summarize the chat session at the user device 120. In another example, a server 100 or user device 120 may generate a specialized lexicon and summarize the chat session.

In step 1230, the device 300 may summarize the contents of the chat session data using the core keyword selected in step 1220. For example, the device 300 may extract "B" as a key keyword corresponding to the "model" subject in the chat session data at step 1220. In addition, the device 300 may extract "update" as a key keyword corresponding to the "problem" In addition, the device 300 may extract "reset" as a key keyword corresponding to the "solution" In this case, the device 300 can summarize the contents of the chat session data in the form of "It is proposed to reset by resetting the B model due to an update-related problem ". Without being limited thereto, the device 300 may summarize the chat session in various forms using the core keyword.

In one embodiment, the device 300 may summarize the content of the chat session data into a table of topics and corresponding key keywords. In another embodiment, the device 300 may summarize the chat session by extracting and listing the sentences containing the key keywords extracted in step 1220 from the chat session. The result of summarizing the chat session may be provided to the user device 120 after the chat session of the user device 120 and the external device 140 ends. Alternatively, when the user device 120 and the external device 140 start a chat, a result summarizing the contents of the previous chat session data may be provided to the user device 120 or the external device 140. [

13 is a diagram illustrating a chat session in accordance with one embodiment. One chat session may include chat content from the time the chat between the user device 120 and the external device 140 starts until the chat session ends. 13, the chat session may include chat content from the chat window 1310 to the chat window 1350. [ In one embodiment, the chat data in step 1110 of FIG. 11 may be a keyword contained in one sentence included in the chat session. In another embodiment, the chat data may include keywords contained in one or more chat windows 1310-1350 included in the chat session, or keywords contained throughout the chat session. In one embodiment, the device 300 may analyze the emotion of the user of the external device 140 that appeared in one chat session after the chat of the user device 120 and the external device 140 is terminated. In another embodiment, the device 300 may analyze the emotion of the user of the external device 140 while the user device 120 and the external device 140 are chatting. For example, the device 300 may analyze the emotion of the user of the external device 140 appearing in each of the sentences included in the chat data. Alternatively, the device 300 may analyze the emotion of the user of the external device 140 appearing in each of the chat windows included in the chat data.

14 is a flow chart showing more specifically step 1110 in Fig. 11 or step 1210 in Fig. In step 1410, the device 300 may obtain chat data indicating the result of the user device 120 and the external device 140 performing the chat.

In step 1420, the device 300 may remove the greeting of the chat data obtained in step 1410. [ Greeting is included in most chat data, but it does not have any special meaning and can be removed for accuracy of analysis. The method of removing the greeting can be performed using the preset greeting wordbook. For example, a greeting wordbook may include greetings such as "Hello" or "Thank you." The device 300 can avoid unnecessary analysis by removing the greetings included in the greeting words in the chat data.

In step 1430, the device 300 may separate the chat data from which the greeting is removed in step 1420 on a keyword-by-keyword basis. The method by which the device 300 separates chat data into keywords may be performed using an n-gram method. The device 300 can filter unnecessary keywords by filtering the keywords acquired using the n-gram method. For example, the device 300 may remove keywords that do not have a special meaning, such as articles, articles, quotations, idioms, and the like. The device 300 can remove unnecessary keywords using a predetermined filtering list in which keywords having no special meaning such as article, definite article, survey, idiomatic phrase, etc. are arranged. For example, keywords common to all corpora may be unnecessary keywords. Thus, the device 300 can remove keywords that appear in all the corpus.

The device 300 can acquire a chat keyword that is a result of processing the chat data in steps 1410 to 1430. The device 300 can determine the emotion of the user of the external device 140 in steps 1120 and 1130 of FIG. 11 using the chat keyword. Alternatively, the device 300 can summarize the content of the chat performed by the user device 120 and the external device 140 in steps 1220 and 1230 of FIG. 12 using the chat keyword.

15 is a diagram specifically showing steps 1120 and 1130 of FIG. At step 1510, the device 300 may obtain a keyword. At this time, the obtained keyword may be a keyword separated in step 1110 of FIG. In step 1520, the device 300 may calculate emotional scores for each of the keywords obtained in step 1510 using the emotional wordbook.

In step 1530, the device 300 may process the irony using the emotion score for each of the keywords calculated in step 1520. For example, the device 300 can determine whether the keyword itself is positive, but uses the irony in the context. In one embodiment, the device 300 may compare emotional scores of keywords included in the same sentence or series of sentences. For example, all other keywords in the same sentence or series of sentences show negative emotions, and one other keyword may represent a positive emotion. In this case, the device 300 can determine that the keyword indicating the positive emotion is not the positive emotion but the keyword used in the irony method. In this case, the device 300 can determine the emotion of the user of the external device 140 more accurately by not using the keyword used for the analysis in the analysis.

In step 1540, the device 300 may update the emotion score of each of the keywords. When the user device 120 and the external device 140 perform chatting in English, the user of the external device 140 may input the keyword in capital letters to emphasize the emotion expressed by the keyword. Therefore, when the device 300 calculates the emotion score for the keyword, a predetermined additive point can be given to the emotion expressed by the keyword if a plurality of characters included in the keyword are composed of capital letters. For example, the "Bad" keyword could have an emotional score of -0.6. If the device 300 has acquired the keyword "BAD ", the device 300 can add 0.2 point to the keyword" BAD " to determine that it has a sensitivity score of -0.8. In another embodiment, the device 300 may assign a given point of addition to the emotional score even when the keyword is used with an exclamation point (!).

At step 1550, the device 300 may determine the emotion of the user of the external device 140. For example, the device 300 can determine the emotion of the user of the external device 140 appearing in the chat data by summing emotional scores corresponding to the keywords included in the chat data. The device 300 can determine that the user of the external device 140 shows a positive emotion when the sum of the emotion scores is positive. In addition, the device 300 can determine that the user of the external device 140 shows a negative emotion when the sum of the emotion scores is negative. The device 300 can determine that the user of the external device 140 shows stronger emotions as the absolute value of the sum of the emotion scores is larger. The device 300 can determine that the user of the external device 140 indicates a neutral sensibility if the absolute value of the sum of the sensibility scores does not exceed the predetermined reference value.

In one embodiment, the device 300 is able to observe in real time the transition in which the user's emotions change. For example, the device 300 may update the emotion of the user based on the received keyword while performing a chat with the user. The device 300 can record the change of emotion. For example, the device 300 can observe and record a transition in which a user's emotions change in a positive or negative direction.

In addition, the device 300 can extract non-verbal emotions of the user. In one embodiment, the device 300 may update the emotion of the user based on the user's chat rate. For example, the device 300 may determine that the user's emotional response is negative if the user's chatting speed is getting faster or slower.

One embodiment may also be implemented in the form of a recording medium including instructions executable by a computer, such as program modules, being executed by a computer. Computer readable media can be any available media that can be accessed by a computer, and can include both volatile and nonvolatile media such as RAM, removable and non-removable media. In addition, the computer-readable medium can include both computer storage media and communication media. Computer storage media may include both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Communication media typically includes computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave, or other transport mechanism, and may include any information delivery media. For example, computer storage media may be embodied in ROM, RAM, flash memory, CD, DVD, magnetic disk, magnetic tape, or the like. According to one embodiment, a method is provided for a device 300 to analyze emotions, the method comprising: generating a first keyword group including at least one keyword indicative of positive emotions and a second keyword including at least one keyword indicative of negative emotions Collecting social network data including a third keyword that is not included in the first emotional wordbook from the social network; and, in the social network data, Based on a frequency that coincides with at least one keyword included in the keyword group and a frequency that the third keyword appears in common with at least one keyword included in the second keyword group, In order to execute the step of generating the second emotional wordbook by adding the keyword The program can be recorded.

100: Server
120: User device
140: External device

Claims (25)

In a method for a device to analyze sensibility,
Obtaining a first emotional word list including a first keyword group including at least one keyword indicating positive emotion and a second keyword group including at least one keyword representing negative emotion;
Collecting social network data including a third keyword not included in the first emotional wordbook from the social network;
In the social network data, a frequency at which the third keyword co-occurs with at least one keyword included in the first keyword group and a frequency at which the third keyword is included in the second keyword group Generating a second emotional wordbook by adding the third keyword to the first keyword group or the second keyword group based on a frequency that coincides with the keyword of the first emoticon wordbook; / RTI > The method according to claim 1,
Wherein the step of generating the second emoticon wordbook comprises:
The third keyword is added to the first keyword group when the frequency with which the third keyword appears in common with at least one keyword included in the first keyword group exceeds a predetermined reference value in the social network data ,
In the social network data, adding the third keyword to the second keyword group when the frequency with which the third keyword appears in common with at least one keyword included in the second keyword group exceeds a predetermined reference value ≪ / RTI > 3. The method of claim 2,
The first emotional word list includes:
Further comprising a third keyword group including at least one keyword indicating a neutral sensibility,
Wherein the step of generating the second emoticon wordbook comprises:
The third keyword is added to the third keyword group when the frequency with which the third keyword appears in common with at least one keyword included in the third keyword group exceeds a predetermined reference value in the social network data ≪ / RTI > The method according to claim 1,
The method comprises:
Receiving chat data from an external device;
Determining emotions corresponding to the keywords included in the chat data using the second emotional wordbook; And
Determining emotion of the user of the external device based on emotions corresponding to each of the keywords included in the chat data; ≪ / RTI > 5. The method of claim 4,
Wherein the first emotional wordbook and the second emotional wordbook include emotional scores obtained by quantifying the degree of emotional corresponding to at least one keyword included in the first keyword group or the second keyword group,
Wherein the step of determining emotion comprises:
And determining the emotional score for at least one keyword included in the chat data. 6. The method of claim 5,
Wherein the step of determining emotion comprises:
Determining the emotion based on a sum of emotion scores for one or more keywords included in the chat data. 5. The method of claim 4,
Further comprising changing a chatting mode with the user based on the determined emotion. A method for a device to summarize chat content,
Comprising: obtaining a first specialized linguistic dictionary including a first group of keywords including at least one keyword associated with a first subject and a second group of keywords including at least one keyword associated with a second subject; And
Wherein a frequency in which a keyword that is not included in the first specialized lexicon is included in at least one keyword included in the first keyword group and a frequency in which at least one keyword included in the second keyword group Generating a second specialized wordbook by adding a keyword that is not included in the first specialized wordbook to the first keyword group or the second keyword group based on a frequency that appears in common with the first specialized wordbook; / RTI > 9. The method of claim 8,
Obtaining chat session data indicating a result of the chat performed with the external device;
Extracting at least one core keyword from the keywords included in the chat session data using the second specialized lexicon; And
Summarizing the contents of the chat session data using the at least one core keyword; ≪ / RTI > 10. The method of claim 9,
The extracting of the key keyword may include:
And extracting keywords included in the second specialized lexicon among the keywords included in the chat session data. 10. The method of claim 9,
Summarizing the above contents,
Summarized into sentences containing the core keyword,
And summarizing the core keyword and the subject corresponding to the core keyword into a table matched to each other. The method according to claim 1 or 8,
Wherein the step based on the common occurrence frequency comprises:
Using pointwise mutual information (PMI). In a device for analyzing emotion,
Obtaining a first emotional word list including a first keyword group including at least one keyword indicating positive emotions and at least one keyword representing negative emotions from the social network, A communication unit for collecting social network data including a third keyword not included in the wordbook;
A frequency in which the third keyword appears in common with at least one keyword included in the first keyword group and a frequency in which the third keyword appears in common with at least one keyword included in the second keyword group, A control unit for generating the second emotional wordbook by adding the third keyword to the first keyword group or the second keyword group based on the frequency of occurrence of the second emoticon wordbook; And
A storage unit for storing the first emotional wordbook and the second emoticonbook; . 14. The method of claim 13,
Wherein,
When the second emotional wordbook is generated,
The third keyword is added to the first keyword group when the frequency with which the third keyword appears in common with at least one keyword included in the first keyword group exceeds a predetermined reference value in the social network data ;
In the social network data, adding the third keyword to the second keyword group when the frequency with which the third keyword appears in common with at least one keyword included in the second keyword group exceeds a predetermined reference value , device. 15. The method of claim 14,
The first emotional word list includes:
Further comprising a third keyword group including at least one keyword indicating a neutral sensibility,
Wherein,
When the second emotional wordbook is generated,
The third keyword is added to the third keyword group when the frequency with which the third keyword appears in common with at least one keyword included in the third keyword group exceeds a predetermined reference value in the social network data , device. 14. The method of claim 13,
Wherein,
Receive chat data from an external device;
Wherein,
Determine emotions corresponding to the keywords included in the chat data using the second emotional wordbook;
And determines the emotion of the user of the external device based on emotions corresponding to each of the keywords included in the chat data. 17. The method of claim 16,
Wherein the first emotional wordbook and the second emotional wordbook include emotional scores obtained by quantifying the degree of emotions corresponding to at least one keyword included in the first keyword group or the second keyword group,
Wherein,
And determines the emotional score for at least one keyword included in the chat data. 18. The method of claim 17,
Wherein,
And determines the emotion based on a sum of emotion scores for one or more keywords included in the chat data. 17. The method of claim 16,
Wherein,
And changes the chatting method with the user based on the determined emotion. A device for summarizing chat content,
Obtaining a first specialized vocabulary field comprising a first group of keywords including at least one keyword associated with a first subject and a second group of keywords including at least one keyword associated with a second subject;
Wherein a frequency in which a keyword that is not included in the first specialized lexicon is included in at least one keyword included in the first keyword group and a frequency in which at least one keyword included in the second keyword group A control unit for generating a second specialized wordbook by adding a keyword that is not included in the first specialized wordbook to the first keyword group or the second keyword group based on a frequency that appears in common with the first specialized wordbook; And
A storage unit for storing the first specialized wordbook and the second specialized wordbook; . 21. The method of claim 20,
Wherein,
Acquiring chat session data indicating a result of the chat performed with the external device;
Extract one or more core keywords from the keywords included in the chat session data using the second specialized lexicon;
And summarizes the contents of the chat session data using the one or more core keywords. 22. The method of claim 21,
Wherein,
When extracting the core keyword,
And extracts keywords included in the second specialized lexicon from among the keywords included in the chat session data. 22. The method of claim 21,
Wherein,
In summary,
Summarized into sentences containing the core keyword,
And summarizing the key keyword and the subject corresponding to the key keyword into a table matched with each other. The method according to claim 13 or 20,
Wherein,
Based on the frequency of common appearance,
RTI ID = 0.0 > PMI. ≪ / RTI > A computer-readable recording medium recording a program for causing a computer to execute the method of claim 1.

Images