KR101812933B1 - An electronic spokesman system - Google Patents

Abstract

An electronic spoken system according to an embodiment of the present invention includes a search unit for searching for a publication document associated with a user at an online site by crawling through a natural language processing (NLP) technique, A determination unit for extracting keywords included in a set keyword list and determining an emotional state of the posted document according to the extracted keywords, a determination unit for determining correspondence information extracted according to the emotional state as a response document for the posted document, A communication unit for transmitting the posting document and the corresponding information to a terminal registered by the user and receiving feedback information on the corresponding information from the terminal, And an updating unit for generating and updating the user keyword information It is gong.

Description

An electronic spoken system {AN ELECTRONIC SPOKESMAN SYSTEM}

The present invention relates to an electronic spoken person system, and more particularly, it relates to an electronic spoken person system, which searches for a user-related publication document that occurs on-line and judges the emotion state, and stores correspondence information corresponding to the emotion state into a response document To an electronic spokesperson system for automatically posting.

Recently, various kinds of information are transmitted through media such as newspapers, broadcasts, magazines, and the Internet. These media are domestically produced and transmitted through the political, economic, social and cultural aspects of the country.

Among the articles delivered through these media, it includes information on corporate image, promotion of new products, and information on major projects. Among these publicized information and information, both the positive and negative aspects of the business are addressed.

In this way, when the article is positive, various information about the company through newspapers or broadcasts is scrapped or recorded by the public relations department or the management department of the company, and is utilized as a means of re-publicizing the image of the enterprise through offline or online Or is being utilized in marketing or brand management of companies.

Also, as the number of users using social network services such as blogs, Twitter, and Facebook skyrocket, the number of articles expressing feelings about specific events, products, and policies has also increased exponentially. These emotional expressions are one of the most important factors in the strategic planning of companies that plan events or products, or institutions that establish policies.

As social network services become popular on the Internet, opinions of diverse users exist. Especially, when malicious exploits or misinformations are displayed on individuals, businesses, institutions, etc., individuals, The Internet and reality are becoming more and more equal so that defamation and product sales can be fatal.

However, honor or damage that has been compromised on the Internet is a difficulty for the victim or the victim to refute by calling himself or herself, In particular, since damage to the voices is weakened due to low impact power and subscription power, the damage to individuals, enterprises, and institutions is gradually increased every time the public is lost interest. And the electronic spokesperson system automatically responds to the posted electronic spokesperson system.

1. Registration No. 10-1161245 (June 25, 2012) 2. Open Patent No. 10-2009-0048997 (2009.05.15)

It is an object of the present invention to provide an electronic spoken system capable of minimizing the damage caused by malicious writing on-line and reducing the time and effort consumed in responding to malicious texts.

Another object of the present invention is to enable quick response to repetitive misreporting and non-repetition, because the emotion state of the posted document can be automatically responded based on the user's keyword.

It is still another object of the present invention to provide a method and apparatus for generating a sentence index for each keyword, even if the posted keywords in the posted document related to the user are all interspersed with keywords corresponding to the affirmative, So that it can judge the state.

Yet another object of the present invention is to provide a method and apparatus for receiving correspondence information from a user when the emotion state of a publication document related to the user is in a neutral state, .

According to an aspect of the present invention, there is provided an electronic spokesperson system comprising: an electronic spokesperson system for crawling a publication document associated with a user at an online site through a natural language processing (NLP) A determination unit for extracting keywords included in a predetermined keyword list from the publication document and determining an emotion state of the posted document according to the extracted keywords, A response unit for posting correspondence information on the site as a response document to the posted document, a communication unit for transmitting the posted document and the correspondence information to a terminal registered by the user and receiving feedback information on the correspondence information from the terminal, According to the feedback information, the keywords included in the publication document are converted into user keyword information And it characterized in that it comprises updating portion updates.

According to an embodiment of the present invention, the response unit automatically posts the response document without judging the emotion state when a predetermined number or more of the keywords included in the user keyword information are extracted from the posted document .

According to an embodiment of the present invention, the determination unit may determine the keywords as a first keyword corresponding to the first emotion state, a second keyword corresponding to the second emotion state, a third keyword corresponding to the third emotion state Counting the number of each keyword, calculating an emotion index (E) of the emotion state through the emotion index calculating formula applied according to the first and second keywords, and when the emotion index is a positive number, State and a neutral state, and when the emotion index is a negative number, it is determined to be in an unauthorized state.

According to an embodiment of the present invention,

이고,

ego,

Where A is a positive constant applied according to the difference between the first and second number of keywords, A is the average of the first weights previously set for each of the first keywords, B is the number of hits 2 is a negative constant applied according to the difference between the first and second number of keywords, C is an average of second predetermined weights for each second keyword, N1 is the number of the first keywords, D is the second weight, N2 is the number of the second keywords, and N3 is the total number of the first and second keywords.

According to an embodiment of the present invention, the determining unit may determine that the posted document is the first emotional state when the emotion index is a value of 1 or more, and when the emotion index is a value of 0 or more and less than 1, Judges the document as the third emotional state, and judges the posted document as the second emotional state when the emotional index is a value less than 0, and when the update section judges the third emotional state, And transmits the correspondence information of the posted document to the terminal, and updates the user keyword information according to whether or not the user is allowed to receive the broadcast transmitted from the user.

The electronic spoken system according to at least one embodiment of the present invention configured as described above minimizes the damage caused by malicious writing that occurs on-line and reduces the time and effort spent in responding to malicious writing There is an effect.

In addition, since the emotional state of the posted document can be automatically responded based on the user's keyword, there is an effect that prompt response to repetitive misreporting or non-verbal sentence is possible.

Also, even if the posted keywords in the posted document related to the user are all interspersed with the keywords corresponding to the affirmative state, the irregular state, and the neutral state, it is possible to calculate the emotion index for each keyword, .

Also, when the emotional state of the posted document related to the user is in the neutral state, it is possible to receive the correspondence information from the user and to update the keyword information of the user separately to provide customized correspondence information for each user.

1 is an electronic spokesperson system diagram according to an embodiment of the present invention.
2 is a block diagram of an electronic spoken language server according to an exemplary embodiment of the present invention.
3 is an operational flow diagram of an electronic spokesman system.

Hereinafter, an electronic spoken system according to the present invention will be described in detail with reference to the drawings. The suffix "module" and " part "for the components used in the following description are given or mixed in consideration of ease of specification, and do not have their own meaning or role. In the present specification, the same or similar reference numerals are given to different embodiments in the same or similar configurations. As used herein, the singular forms "a", "an" and "the" include plural referents unless the context clearly dictates otherwise.

It is to be understood that the drawings herein illustrate conceptual aspects of exemplary circuits embodying the principles of the invention. That is, it should be understood that the functions shown in the figures may be substantially represented on a computer-readable medium and performed by various processes performed by the computer or processor, whether or not the computer or processor is explicitly shown.

And, each function can be provided not only with dedicated hardware, but also with the use of hardware capable of executing software in connection with appropriate software. When provided by a processor, the functions may be provided by a single dedicated processor, a single shared processor, or a plurality of individual processors, some of which may be shared.

In the following description, well-known functions or constructions are not described in detail since they would obscure the invention in unnecessary detail. Hereinafter, a preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings.

1 is an electronic spokesperson system diagram according to an embodiment of the present invention.

Referring to FIG. 1, an electronic spoken system 1000 may include a user terminal 10, an electronic spoken server 100, and a plurality of web servers 200-1 to 200-N.

First, the user terminal 10 is a terminal carried by an individual, a company, or a government office. The user terminal 10 is connected to the electronic spokenman server 100 through the network 10 and is implemented as a computer, a portable terminal, or a television .

For example, the computer includes a notebook computer, a desktop computer, a laptop computer, and the like on which a web browser (WEB Browser) is mounted. The portable terminal is, for example, Communication System, GSM (Global System for Mobile communications), PDC (Personal Digital Cellular), PHS (Personal Handyphone System), PDA (Personal Digital Assistant), IMT (International Mobile Telecommunication) -2000, CDMA (Code Division Multiple Access) 2000, a W-CDMA (W-Code Division Multiple Access), a Wibro (Wireless Broadband Internet) terminal, a smart phone and the like. In addition, the television may include an Internet Protocol Television (IPTV), an Internet television (TV), a terrestrial TV, a cable TV, and the like.

The user terminal 10 may also receive a web page such as various HTML (Hyper Text Markup Language) documents provided by the electronic spokesperson server 100 and display it on a conventional web browser A Web browser), and may be any type of wired / wireless communication device capable of accessing the electronic spokesman server 100 and using various Web services.

The network 50 includes a plurality of services, such as HTTP (Hyper Text Transfer Protocol), Telnet, File Transfer Protocol (FTP), Domain Name System (DNS) (Hereinafter, referred to as a " network "), a transfer protocol, a simple network management protocol (SNMP), a network file service (NFS), and a network information service (NIS) And can be connected to an electronic spoken server 100, which will be described later, in an individual, authorized person, company, or government office. Meanwhile, the network 50 may be a wired or wireless Internet, or may be a core network integrated with a wired public network, a wireless mobile communication network, or a portable Internet.

Next, the electronic spokesman server 100 crawls through a natural language processing (NLP) technique in cooperation with a server providing a pre-registered online site or a social network service from the user terminal 10 , A publication document associated with a user, a web document, a publication, and the like (hereinafter, referred to as a 'publication document').

That is, the publication document associated with the user may be a post including the family of the user, the affiliation of the user, the nickname of the user, the product name of the user, and the business name based on the user information, e.g., the user name and the user ID.

Here, the term "natural language processing" (NLP) refers to a process of recognizing and processing a language commonly used by humans on a computer, and may include information retrieval, query response, system automatic translation, and interpretation. At this time, natural language is a term distinguished from 'machine language' which is a machine-understandable language, and it means 'a language that can be understood by humans (Korean, English, Japanese, etc.)'.

For example, the electronic spokesperson server 100 may use Levenshtein Distance, Pointwise Mutual Information (PMI), Support Vector Machines (SVMs), and Naive ≪ / RTI > Bayes). ≪ / RTI >

Also, the term " crawling " may refer to a technique for searching for a publication document associated with a user, with all the title, content, and date of the publication document as search targets.

2, the configuration of the electronic spokesman server 100 will be described in more detail.

Next, the plurality of web servers 200-1 to 200-N are external servers on which posted documents related to users generated in the online sites are displayed, and are specific web sites (such as a text site, a broadcasting site, a personal site, A blog, a cafe, or an external server having a site supporting a social network service.

2 is a block diagram of an electronic spoken language server according to an exemplary embodiment of the present invention.

Referring to FIG. 2, the electronic spokesman server 100 may include a search unit 110, a determination unit 120, a response unit 130, a communication unit 140, and an update unit 150.

First, the search unit 110 crawls a publication document associated with a user at an online site of a plurality of web servers 200-1 to 200-N through natural language processing (NLP) can do.

Next, the determination unit 120 determines whether or not a keyword included in a predetermined keyword list from the publication document that is posted on the online sites of the plurality of web servers 200-1 to 200-N and is searched through the search unit 110 And may determine an emotion state (Opinion) for the posted document according to the extracted keywords.

Here, the keyword means words selected manually or words manually input from the user, depending on the emotional state, among the words to be posted on the posted document such as articles, reviews, and posts, and the emotional state may be an individual, , Government office, and product, and may be any one of positive, negative, and neutral.

That is, the judging unit 120 judges whether the publication document is a review for a specific product, a slander about a specific person, an obligation for a specific company, a review, a slander, It is possible to determine the emotional state for one of the positive state, the negative state, and the neutral state.

For example, when the posted document is a non-repudiation directed to a particular person, the determination unit 120 may determine that the emotion state of the posted document is an unauthorized state according to the extracted keywords.

According to the embodiment, the electronic spokesman server 100 further includes a storage unit 160. The storage unit 160 stores the keywords obtained in advance in the emotion state, for example, the keywords classified according to the affirmative state, It can be saved in advance as a list.

That is, the storage unit 160 stores the affirmative state keyword list including the keywords corresponding to the affirmative state of the emotional state, the irregular keyword list including the keywords whose emotional state corresponds to the negative state, The neutral state keyword list including the keywords can be classified and stored in advance.

Here, the keyword list is an example of a list of affirmative state keywords in the keyword list, and Table 2 below is an example of the affirmative state keyword list in the following Table 1, which is a list including the respective keywords classified according to the emotional state and the predetermined first weights according to the respective keywords. Table 3 is an example of a neutral status keyword list.

Positive status keyword list The first weight (1 to 2) Good 1.8 Jackpot 1.6 Recognition 1.3 awesome 1.5 .
.
. .
.
.

Negative keyword list The first weight (1 to 2) Dog Oba 1.7 Truth 1.5 Push 1.6 worst 1.3 .
.
. .
.
.

Neutral status keyword list The first weight (1) Unkindness One situation News joke .
.
.

That is, the judging unit 120 judges the keywords extracted from the posted document through the keyword list stored in advance in the storage unit 160 as a first keyword corresponding to a first emotion state, for example, a first keyword corresponding to an affirmative state, A second keyword corresponding to an unfixed state, and a third keyword corresponding to a third emotion state, for example, a neutral state, and count the number of keywords in each emotion state.

In addition, the determination unit 120 calculates the emotion index E of the emotion state for the corresponding publication document through the emotion index calculation formula applied according to the first and second keywords among the keywords of the counted emotion states, It can be determined that the emotion index is negative or positive when the emotion index is positive and when the emotion index is negative.

Here,

(1) 이고,

(1)

Wherein A is a positive constant applied according to the difference in the number of the first and second keywords, A is an average of predetermined first weights for each of the first keywords, B is a number 2 is a negative constant applied according to the difference between the first and second number of keywords, C is an average of second predetermined weights for each second keyword, N1 is the number of the first keywords, D is the second weight, N2 is the number of the second keyword, and N3 is the total number of the first and second keywords.

That is, the storage unit 160 stores the number of the first and second keywords according to the difference between the second weight (1 to 2) according to the number of views of the posted document and the number of the first and second keywords The applied positive (+) or negative (-) constants can be stored in advance.

First site (1 million members) Second site (10,000 members) ... views The second weight views The second weight ... Less than 100 times 1.1 Less than 50 times 1.1 ... 101 times to 300 times 1.2 51 times to 80 times 1.2 ... 301 times - 400 times intervals 1.3 81 times - 110 times intervals 1.3 ... .
.
. .
.
. .
.
. .
.
. .
.
.

Number difference alpha (1 to 1.2) Number difference beta 0 to 2 segments One -2 to 0 sections -One 3 to 5 sections 1.3 -5 to -3 sections -1.1 Greater than 5 2 Greater than -5 -2

The determination unit 120 may determine that the posted document is an affirmative state that is the first emotion state when the emotion index E of the emotion state for the posted document is equal to or greater than 1, Is equal to or greater than 0 and less than 1, determines that the posted document is a neutral state that is the third emotion state, and when the emotion index (E) is a value less than 0, .

According to the embodiment, when a keyword corresponding to a specific word, for example, a profanity or a specific event, is selected from the predetermined keyword list from the posted document, the determination unit 120 determines whether the emotion state of the posted document is a simple profanity Or a specific event.

Next, the responding unit 130 can automatically post correspondence information extracted according to the emotional state, as a response document to the posted document, on the corresponding site.

Here, the correspondence information may be information previously input from the user, for example, a reply, and may be the automatic response information previously stored in the storage unit 160 from the manager according to the emotion index E. [

For example, when the responding unit 130 is judged as a second emotional state judged to be simple profanity, "a legal action can be taken for unfounded slander and abuse, and oo's articles and information are automatically stored Please note." Or "oo is made from a misunderstanding related to o case, so it has already been called as o, so oo says that the article left is a misunderstanding, and even a defamation." Or "Thank you for your interest in any form." A response document for the same publication document can be automatically posted to the corresponding site where the publication document is posted.

Next, the communication unit 140 transmits the shortcut menu and corresponding information directly linked with the posted document and the corresponding information to the user terminal 10 registered by the user, and transmits the corresponding information from the terminal 10 Feedback information for the UE can be received.

Here, the feedback information may be a rating score of the user for the corresponding information.

Thereafter, the update unit 150 may generate the keyword included in the posted document as user keyword information according to the feedback information, and may update the stored keyword to the storage unit 160.

Here, the user keyword information may be a list including specific words repeatedly posted for each user.

Accordingly, when a predetermined number or more of the keywords included in the user keyword information updated through the updating unit 150 are extracted from the posted document to be searched through the searching unit 110, the determining unit 120 The correspondence information previously responded can be automatically posted to the posted document without the operation of judging the emotional state via the message.

The update unit 150 may update the correspondence information of the posted document through the communication unit 140 to the user terminal 10 ), And update the keywords extracted from the posted document to the user keyword information according to whether or not the user terminal 10 is allowed to receive a posting.

3 is an operational flow diagram of an electronic spokesman system.

1 to 3, the search unit 110 may search for a publication document associated with a user in an online site provided by a plurality of Web servers 200-1 to 200-N (D) using natural language processing (NLP ) Technique (step S100).

At this time, the determination unit 120 extracts the keywords included in the predetermined keyword list from the posted document (S110), and may determine the emotion state of the posted document according to the extracted keywords (S120).

Next, the response unit 130 extracts corresponding information from the storage unit 160 according to the emotion state determined through the determination unit 120, and outputs the corresponding information as a response document to the posted document, (S130). ≪ / RTI >

At this time, the communication unit 140 transmits the posted document and the corresponding information to the user terminal 10 so as to confirm the received information, and receives feedback information on the corresponding information from the user terminal 10 (S140).

Next, the update unit 150 may generate and update the keywords included in the publication document as user keyword information according to the feedback information received from the user terminal 10 through the communication unit 140 (S150).

When a certain number of keywords included in the user keyword information are extracted from other posted documents searched by the search unit 110 (S160), the responding unit 130 may automatically (S130).

That is, the responding unit 130 can extract malicious non-verbal or repetitive misunderstandings by specific groups, which are repeatedly performed according to the user keyword information, on other posted documents retrieved by the search unit 110, It is possible to perform a quick response to the request.

The above-described electronic spokesman system can be applied to a configuration and a method of the embodiments described above in a limited manner, but the embodiments may be modified so that all or some of the embodiments are selectively combined do.

10: User terminal
50: Network
100: Electronic spoken server
110:
120:
130:
140:
150: Update section
200-1 to 200-N: a plurality of web servers
1000: Electronic spokesman system

Claims (5)

A search unit for searching a posted document associated with a user in an online site by crawling through a natural language processing (NLP) technique;
A determining unit for extracting keywords included in a predetermined keyword list from the posted document and determining the emotional state of the posted document according to the extracted keywords;
A response unit for posting corresponding information extracted according to the emotional state on the site as a response document to the posted document;
A communication unit for transmitting the posted document and the corresponding information to a terminal registered by a user and receiving feedback information on the corresponding information from the terminal; And
And an update unit for generating and updating the keywords included in the publication document according to the feedback information as user keyword information.
The apparatus according to claim 1,
And automatically posts the response document without judging the emotion state when a predetermined number or more of the keywords included in the user keyword information are extracted from the posted document.
The apparatus of claim 1,
The keywords are classified into a first keyword corresponding to the first emotion state, a second keyword corresponding to the second emotion state, and a third keyword corresponding to the third emotion state,
The emotion index E of the emotional state is calculated through the emotion index calculating formula applied according to the first and second keywords,
When the emotion index is a positive number, judges to be an affirmative state or a neutral state, and when the emotion index is a negative number, it is judged to be an unauthorized state.
4. The method according to claim 3,

ego,
Where A is a positive constant applied according to the difference between the first and second number of keywords, A is the average of the first weights previously set for each of the first keywords, B is the number of hits 2 is a negative constant applied according to the difference between the first and second number of keywords, C is an average of second predetermined weights for each second keyword, N1 is the number of the first keywords, D is the second weight, N2 is the number of the second keyword, and N3 is the total number of the first and second keywords.
5. The apparatus of claim 4,
Judges that the posted document is in the first emotion state when the emotion index is a value of 1 or more,
Judges that the posted document is in the third emotional state when the emotion index is a value of 0 or more and less than 1,
Judges that the posted document is in the second emotional state when the emotion index is a value less than 0,
Wherein,
And transmits the correspondence information for the publication document to the terminal through the communication unit when the third emotion state is determined, and updates the user keyword information according to whether the publication permission is received from the user. .

Images