KR101491628B1 - Method, apparatus and system for extracting keyword affecting for mood change of the public using blog - Google Patents

Abstract

A method for extracting keywords affecting the emotional change of the public from the blog comprises: an emotion index calculating step of calculating an emotion index using an emotional vocabulary dictionary based on vocabularies included in each comment in the blog; an emotional change section extracting step of extracting an emotional change section of the blog using the calculated emotion index; and a keyword extracting step of collecting comments included in the emotional change section and extracting main keywords included in the emotional change section from the collected comments. Therefore, when keywords affecting the emotional change of the public are extracted, the method helps to infer reasons associated with the emotional change of the public in a specific company or social issue to easily identify the needs of the public, thereby creating economic benefits later.

Description

TECHNICAL FIELD [0001] The present invention relates to a keyword extracting method, apparatus, and system for extracting keywords from a blog,

The present invention relates to a keyword extraction method, apparatus, and system, and more particularly, to a method, apparatus, and system for extracting keywords that affect emotional change through blog analysis.

In the case of microblogging, which seeks convenience, diversity and information sharing among users, it expresses individual opinions freely, so it contains a lot of emotional information, and this emotional information is continuous due to a specific issue, There are changing characteristics. Particularly, as social network service (SNS) such as Twitter or Facebook is popularized, researches using it have been actively conducted.

Conventional emotional analysis methods assume that emotion is static and merely perform polarity analysis with positive or negative content of microblog. Various attempts to analyze the emotion of the document have not yet been conducted, And microblogs, as well as the accuracy of emotional analysis results for documents with limited number of characters. In addition, microblogs contain a lot of emotional information because individual opinions can be freely expressed. Such emotional information is characterized by a constant change due to a specific issue or simple psychological change of an individual. Is not reflected.

An object of the present invention to solve the above-mentioned problems is to provide a method of extracting keywords that affect the emotional change of the public, and the emotional analysis is performed through the emotional dictionary using only the subjective words of the contents of the microblog, An apparatus and a system for extracting a keyword related to a specific topic in consideration of an emotion index varying according to a keyword.

This allows us to identify the factors that affect emotional change for a particular topic and to better understand the needs of the public.

According to another aspect of the present invention, there is provided a method of extracting keywords affecting a change in emotional state of a public, the method including: calculating an emotion index using an emotional vocabulary dictionary based on vocabulary included in each comment in the blog; An emotion index calculating step of extracting emotional change intervals of the blog using the emotion index, and a keyword extracting step of collecting comments of the emotion change interval and extracting main keywords of the emotion change interval can do.

The keyword extraction method may further include collecting at least one blog including a word for a specific topic at each reference time.

The step of calculating the emotion index may include the step of acquiring the refined data by removing a function word that does not serve to represent the contents of a sentence through a data preprocessing process for each comment.

The emotional vocabulary dictionary may be a vocabulary dictionary that quantifies the degree of a positive or negative semantic direction of a specific vocabulary.

The emotion index calculating step may include calculating a positive index and a negative index for the vocabulary in each of the comments using the emotional vocabulary dictionary and subtracting the sum of the negative indices from the sum of the positive indices, Value to the emotion index of each of the comments.

The step of calculating the emotion index may include calculating an emotion index of each of the comments existing within the reference time, and calculating an interval emotion index.

The emotion index calculation step may include calculating the interval emotion index using an average of the emotion indexes of the respective comments existing within the reference time.

The emotional change interval extraction step may include calculating the interval in which the temporal change of the interval emotion index is greatest as the emotional change interval.

The emotional change section may include a step of calculating a section in which the temporal change of the section emotion index is greater than a preset threshold value as the emotion change section.

The keyword extracting step may include extracting a main keyword of the emotion change interval having a high frequency of appearance using a term frequency-inverted document frequency (TF-IDF).

The keyword extracting step may include extracting the main keyword based on an objective vocabulary excluding the vocabulary used for calculating the emotion index among the contents of the comment of the emotion change section.

The blog may be a micro-blog.

The comment may include at least one of a tweet, a reply, and a comment.

According to an embodiment of the present invention, there is provided an apparatus for extracting a keyword having an influence on a change in emotional state of a public, the emotional index being calculated using an emotional vocabulary dictionary based on a vocabulary included in each comment in the blog An emotion index calculating unit for extracting emotion change intervals of comments in the blog using the emotion index; and a keyword extracting unit for collecting comments of the emotion change interval and extracting main keywords of the emotion change interval .

The keyword extracting apparatus may further include a blog collecting unit collecting at least one blog including a word for a specific topic at each reference time.

The emotional vocabulary dictionary may be a vocabulary dictionary that quantifies the degree of a positive or negative semantic direction of a specific vocabulary.

Wherein the emotion index calculating unit comprises a first calculating unit for calculating an affirmative index and an indefinite index for the vocabularies in the respective comments using the emotional vocabulary dictionary and a second calculating unit for calculating a value obtained by subtracting the sum of the negative indices from the sum of the affirmative indices, And a second calculating unit for calculating an emotion index of the comment of the first user.

The emotion index calculator may calculate the interval emotion index using an average of the emotion indexes of the comments included in the reference time.

The emotion change interval extracting unit may calculate the interval in which the temporal change of the interval emotion index is greatest as the emotion change interval.

The keyword extracting unit may extract a main keyword of the emotion change interval having a high appearance frequency by using a term frequency-inverted document frequency (TF-IDF) as a content of a comment of the emotion change period.

According to another aspect of the present invention, there is provided a system for extracting keywords that affect a change in emotional state of a public person, the system including a server for receiving blog related data and emotional vocabulary dictionary related data from a database, Calculating an emotion index based on a vocabulary included in each comment, extracting a sensibility change section of comments in the blog using the emotion index, collecting comments of the sensibility change section, A database for storing data related to the blog, a database for storing the emotional vocabulary dictionary related data, and a comment on the blog, and transmits the main keyword extraction request to the keyword extracting device, If the main keyword is the keyword It may include a user device for receiving from the output device.

According to the method, apparatus, and system for extracting keywords that affect the emotional change of the public in the blog of the present invention, when extracting keywords that are factors affecting the emotional change of the public, It will help to infer the reasons associated with change, which makes it easier to identify public needs ahead of time, and can later create a gross profit.

1 is a block diagram schematically showing a system to which a keyword extraction method according to an embodiment of the present invention can be applied;
FIG. 2 is a flowchart schematically showing a keyword extracting method according to an embodiment of the present invention;
3 is a detailed flowchart specifically showing the emotion index calculating step of the keyword extracting method according to an embodiment of the present invention,
FIG. 4 is a detailed flowchart specifically illustrating a step of extracting an emotion change section of a keyword extracting method according to an embodiment of the present invention;
FIG. 5 is a detailed flowchart illustrating a key keyword extracting step of a keyword extracting method according to an exemplary embodiment of the present invention;
FIG. 6 is a block diagram schematically showing a keyword extracting apparatus according to an embodiment of the present invention;
FIG. 7 is a detailed block diagram specifically showing an emotion index calculating unit of a keyword extracting apparatus according to an embodiment of the present invention;
8 is a detailed block diagram specifically illustrating a sensibility change section extracting unit of a keyword extracting apparatus according to an embodiment of the present invention,
FIG. 9 is a detailed block diagram specifically showing a keyword extracting unit of a keyword extracting apparatus according to an embodiment of the present invention;
FIG. 10 is a graph showing a change in emotion index of a user for a specific topic according to a keyword extracting method according to an embodiment of the present invention;
11 is a table showing top keywords extracted by applying the keyword extraction method according to an embodiment of the present invention.

While the invention is susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and will herein be described in detail.

It should be understood, however, that the invention is not intended to be limited to the particular embodiments, but includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention.

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

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

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

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

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In order to facilitate the understanding of the present invention, the same reference numerals are used for the same constituent elements in the drawings and redundant explanations for the same constituent elements are omitted.

Keyword Extraction System

1 is a block diagram schematically showing a system to which a keyword extraction method according to an embodiment of the present invention can be applied. 1, the keyword extraction system includes a keyword extraction device 10, user devices 20-1, 20-2, ..., 20-N, databases 30-1 and 30-2, .

Referring to FIG. 1, the keyword extracting apparatus 10 extracts keywords that affect the emotional change of the public in a blog. Here, the blog may include a microblog as well as an Internet café. The microblog may include social network services such as Twitter, Facebook, etc., and may be implemented in the form of an application on a smart phone. The keyword extracting apparatus 10 calculates an emotion index for the vocabulary included in each comment in the blog based on the emotional vocabulary dictionary, extracts the emotion change interval using the emotion index, And extract key keywords of the sensibility change section. Here, the comment may include at least one of a tweet, a reply, and a comment. The keyword extracting apparatus 10 can receive the information (for example, data of a comment, a user, a photograph, and the like) in the blog in the blog related database 30-2 and the information related to the emotional vocabulary dictionary, -1). The keyword extracting apparatus 10 may receive a keyword extraction request from the user devices 20-1, 20-2, ..., 20-N. The keyword extraction request may include information for extracting a keyword for a specific topic. In addition, information for extracting a main keyword within a specified period by specifying a period may be included. The keyword extracting apparatus 10 may transmit the extracted keywords to the user devices 20-1, 20-2, ..., 20-N in response to the request.

The user devices 20-1, 20-2, ..., 20-N are client devices that upload various information to the blog. The user devices 20-1, 20-2, ..., 20-N may be various types of devices such as a smart phone, a PC, and a tablet PC. The user devices 20-1, 20-2, ..., 20-N can transmit a keyword extraction request to extract a keyword by specifying a topic or a term to the keyword extraction device. In addition, the extracted keyword can be received from the keyword extracting apparatus 10. [

The databases 30-1 and 30-2 may include the emotional vocabulary dictionary database 30-1 and the blog related database 30-2. As described above, the emotional vocabulary dictionary database 30-1 stores information related to the emotional vocabulary dictionary. Emotional vocabulary dictionary is a vocabulary dictionary that quantifies the degree of positive or negative semantic direction of a vocabulary. For example, "good" can have a positive index of +0.8 in the positive sense and positive index of +1.4 in the positive sense of "fun". Conversely, "bad" can have negative index of -0.8 in negative meaning and negative index of -1.4 in negative meaning "not interesting". In other words, there is a characteristic that the degree of affirmation or negation is quantified rather than being differentiated into positive or negative.

The blog related database 30-2 stores various information in the blog. The blog related database 30-2 can return only blogs related to a specific topic through a search in response to a request from the keyword extracting apparatus 10. [

According to another embodiment of the present invention, the keyword extracting apparatus 10 may include a blog related database 30-2. That is, the keyword extracting apparatus 10 receives blog-related information including comments directly from the user devices 20-1, 20-2, ..., 20-N serving as a server related to the blog, Stored in the database 30-2, and directly performed data search and data analysis using the blog related database 30-2.

How to extract keywords

2 is a flowchart schematically showing a keyword extraction method according to an embodiment of the present invention.

Referring to FIG. 2, a keyword extracting apparatus according to an embodiment of the present invention receives a specific topic (S210). At this time, when receiving the keyword extraction request through the user device, the topic input may be received based on the specific topic related information included in the received request. If the keyword extraction request is not received, a topic can be input directly from the user through the user interface.

Thereafter, the keyword extracting device collects the blogs related to the input specific topic at each reference time (S220). The keyword extracting device makes a search request related to a specific topic entered in the blog related database and collects the blog corresponding to the search request. This can be performed at each reference time. The reference time means a time interval for calculating the sensibility change interval, and can be set to 12 hours. This can be changed via the user interface. The keyword extraction device collects blogs containing words for a particular topic every 12 hours.

Next, the keyword extracting apparatus calculates an emotion index based on the vocabulary included in the collected comment in the blog (S230). At this time, the emotional vocabulary dictionary 30-1 can be used. That is, the emotional index of comments existing within a specific reference time can be calculated using the emotional vocabulary dictionary 30-1 including the quantified numerical information of the words expressing subjective feelings. The keyword extracting apparatus calculates the average of the interval sensitivity index of the microblog for a specific topic in the corresponding period by using the emotion index for each comment.

Then, the keyword extraction unit extracts the emotion change interval based on the calculated interval emotion index (S240). Based on the calculated interval sensitivity index, a section in which the variation is largest or a section in which the variation is greater than a preset threshold value is extracted as a sensitivity variation section.

Finally, the keyword extraction device analyzes the content of the micro blog based on the extracted point of view of the sensibility change period to extract the main keyword (S250). At this time, the frequency of words is calculated using the TF-IDF (Term Frequency-Inverse Document Frequency), and N keywords having the highest value can be extracted.

FIG. 3 is a detailed flowchart illustrating the emotion index calculating step of the keyword extracting method according to an embodiment of the present invention.

Referring to FIG. 3, in order to calculate the emotion index, the keyword extracting apparatus first collects a blog about a specific topic (S310). This can be done by collecting a microblog containing a word for a particular topic every reference time (e.g., 12 hours). In operation S320, the keyword extracting apparatus analyzes the word of the blog comment collected for the specific topic within the reference time to determine whether it is a word indicating the contents of the sentence. This is to acquire refined data by eliminating unnecessary words such as abbreviations and URLs, which are function words that do not play a big role in expressing the contents of the sentences through the data preprocessing process in order to perform a more accurate emotional analysis. A judgment as to whether or not the contents of a sentence is to be made can be made using an abbreviated word dictionary for the processing of an abbreviated word. As a result of the determination, excluded words and URLs that do not indicate the contents of the sentence are excluded (S330). Then, an affirmative index and a negative index are calculated for the words indicating the contents of the sentence (S340). This is done using the emotional vocabulary dictionary, as described above. The positive and negative indices are calculated using the emotional vocabulary dictionary for each vocabulary in each comment. Then, an affirmation index for the comment is calculated using the positive and negative indices (S350). This can be expressed by the following equation.

Here, Score means an emotion index, t _i means each tweet, w means a word, Posscore means an affirmative index, and Negscore means a negative index. That is, the emotion index for a specific tweet can be expressed by subtracting the sum of negative indices from the sum of affirmative indices of the words included in the tweet.

Then, the section emotion index can be calculated using the average of the emotion index based on the emotion index of each comment included in the reference time (S360).

FIG. 4 is a detailed flowchart illustrating a step of extracting an emotion change section of a keyword extraction method according to an embodiment of the present invention.

Referring to FIG. 4, the keyword extracting apparatus may receive an interval emotion index calculated on a reference time basis in the emotion index calculating step (S410). The keyword extracting apparatus calculates the amount of change of the emotion index using the difference between the previous section emotion index and the current section emotion index based on the received section emotion index, and compares it with the threshold value (S420). At this time, if the change amount is smaller than the threshold value, the current interval is extracted as the emotion change interval (S430). The threshold value can be set in advance and can be changed through the user interface.

According to another embodiment of the present invention, a section having the largest amount of change in the amount of change of the emotion index may be extracted as the emotion change section. At this time, the section to be judged should be appropriately set.

FIG. 5 is a detailed flowchart illustrating key keyword extraction steps of a keyword extraction method according to an exemplary embodiment of the present invention.

Referring to FIG. 5, when a sensibility change interval is determined, a comment in a sensibility change interval is collected (S510). Then, the collected vocabularies are collected (S520). At this time, it is determined whether the word is a word containing an objection meaning (S530). As a result of the determination, if the keyword is not an objective word, it is necessary to extract keywords that affect the emotional change, not the vocabulary representing the emotional word. Therefore, subjective words used in the emotional analysis are excluded in the emotional index calculation step (S540). The TF-IDF is applied based on objective words to extract a plurality of keywords related to the corresponding time point (S550). TF is the number of occurrences of a word in a document at word frequency. This can be calculated as "the number of occurrences of the word in one document / the number of occurrences of all occurrences in a document". The IDF indicates how many of the entire document a word has appeared in. The IDF can be obtained as "the total number of documents / the number of documents in which the word appears". In the present invention, a document may correspond to a comment including tweets, replies, and comments. With TF and IDF, TF-IDF can be calculated as the product of TF and IDF. Finally, the appearance frequency is calculated through the TF-IDF, and the top N keywords among the extracted keywords are extracted as the main keywords (S560). At this time, the number N of extracted keywords can be changed by the user through setting.

Keyword Extractor

6 is a block diagram schematically showing a keyword extracting apparatus according to an embodiment of the present invention. 6, the keyword extracting apparatus according to an exemplary embodiment of the present invention includes a blog collecting unit 610, an emotion index calculating unit 620, a sensibility change period extracting unit 630, and a keyword extracting unit 640, . ≪ / RTI >

Referring to FIG. 6, the blog collection unit 610 receives a specific topic, and collects blogs related to a specific topic input for each reference time. At this time, when receiving the keyword extraction request through the user device, the user can receive the topic input based on the specific topic related information included in the received request, and in some cases, receive the topic directly from the user through the user interface have. The blog collecting unit 610 makes a search request related to a specific topic entered in the blog related database and collects the blog corresponding to the search request. This can be performed at each reference time. The reference time means a time interval for calculating the sensibility change interval, and can be set to 12 hours. This can be changed via the user interface. The keyword extraction device collects blogs containing words for a particular topic every 12 hours.

The emotion index calculating unit 620 calculates the emotion index based on the vocabulary included in the comments in the collected blog. At this time, the emotional vocabulary dictionary can be used. That is, the emotional index of comments existing within a specific reference time can be calculated using the emotional vocabulary dictionary including the quantified numerical information of the words expressing subjective emotions. The emotion index calculating unit 620 may calculate the interval emotion index by calculating the average of the interval sensitivity index of the micro blog for the specific topic in the corresponding period using the emotion index for each comment.

Then, the sensibility change section extractor 630 extracts the sensibility change section based on the calculated emotion index of each section. Based on the computed emotion index of the interval, the interval where the change occurs most greatly or the interval where the change is larger than the preset threshold value is extracted as the emotion change interval.

Finally, the keyword extracting unit 640 analyzes the content of the microblog based on the extracted emotion change interval time point to extract the main keyword. At this time, the frequency of words is calculated using the TF-IDF (Term Frequency-Inverse Document Frequency), and N keywords having the highest value can be extracted.

7 is a detailed block diagram specifically showing an emotion index calculating unit 620 of a keyword extracting apparatus according to an embodiment of the present invention. 7, the emotion index calculating unit 620 according to an embodiment of the present invention includes a lexical index calculating unit 710, a first summing unit 720, a second summing unit 722, A sensitivity index calculating unit 730 and a section sensitivity index calculating unit 740. [

Referring to FIG. 7, the lexical index calculation unit 710 analyzes words of blog comments collected for a specific topic within a reference time, and determines whether the words are words representing the contents of a sentence. This is to acquire refined data by eliminating unnecessary words such as abbreviations and URLs, which are function words that do not play a big role in expressing the contents of the sentences through the data preprocessing process in order to perform a more accurate emotional analysis. A judgment as to whether or not the contents of a sentence is to be made can be made using an abbreviated word dictionary for the processing of an abbreviated word. As a result of judgment, abbreviations and URLs that do not indicate the contents of sentences are excluded. Then, the positive index and the negative index are calculated for the words indicating the contents of the sentence. This is done using the emotional vocabulary dictionary, as described above.

Then, the first summation unit 720 calculates the sum of the affirmative indices by summing the affirmative indices of the respective words, and the second summation unit 722 calculates the sum of negative indices by summing the indefinite indices of the respective words . Then, the per-emotion sensitivity index calculating unit 730 calculates the emotion index for each comment by subtracting the sum of the indefinite exponent from the sum of the affirmative indices.

The interval sensitivity index calculating unit 740 may calculate the interval feeling index using the average of the emotion indexes based on the emotion indexes of the respective comments included in the reference time.

8 is a detailed block diagram specifically showing a sensibility change section extraction unit 630 of the keyword extraction apparatus according to an embodiment of the present invention. 8, the sensibility change section extraction unit 630 may include an emotion index reception unit 810 and a determination unit 820. [

Referring to FIG. 8, the emotion index receiving unit 810 may receive an interval emotion index calculated on a reference time basis from the emotion index calculating unit 620. [ The determination unit 820 calculates the amount of change of the emotion index using the difference between the previous section emotion index and the current section emotion index based on the received section emotion index. Then, it compares it with the threshold value. At this time, when the amount of change is smaller than the threshold value, the current interval is extracted as the emotion change interval if the change amount is larger than the threshold value, without regard to the emotion change interval. The threshold value can be set in advance and can be changed through the user interface.

According to another embodiment of the present invention, the determination unit 820 may extract a section having the largest amount of change in the amount of change of the emotion index as an emotion change period. At this time, the section to be judged should be appropriately set.

FIG. 9 is a detailed block diagram specifically showing a keyword extracting unit 640 of a keyword extracting apparatus according to an embodiment of the present invention. 9, the keyword extracting unit 640 may include a vocabulary extractor 910, a TF-IDF applying unit 920, and an upper keyword extracting unit 930.

Referring to FIG. 9, the vocabulary extractor 910 collects comments in the determined emotion change interval. It then collects the vocabularies in the collected comments. At this time, it is judged whether or not it is a word containing an objective meaning, and if it is not an objective word, it is judged that it is not a keyword which causes a change in sensitivity, and is excluded.

The TF-IDF application unit 920 extracts a plurality of keywords related to the corresponding point by applying TF-IDF based on objective words obtained by subtracting words excluded from the vocabulary-making unit 910.

Finally, the high-level keyword extracting unit 930 calculates an appearance frequency through the TF-IDF applying unit 920 and then extracts the top N keywords among the extracted keywords as a main keyword (S560). At this time, the number N of extracted keywords can be changed by the user through setting.

Simulation result

(1) Experimental data

The data used is a Twitter tweet that is a type of microblog, and was collected for a total of 25 days from March 21, 2011 to April 15, 2011 using Spinn3r API2). After selecting topics as 'Apple' and 'Samsung', tweets containing related words were collected in 12-hour increments, and the contents of the tweets and the time recorded in the tweets were used. 606,357 for 'Apple', and 41,715 for 'Samsung'. We eliminated unnecessary factors such as stopwords and URLs, which are functional words that do not play a big role in expressing contents in the sentences through preprocessing of the collected data.

(2) Experimental results

Twenty - five days' tweets including 'Apple' and 'Samsung' designated as a specific topic were collected and emotional analysis was performed in 12 - hour unit using SentiWordNet emotion dictionary.

FIG. 10 is a graph illustrating changes in emotion index of a user to a specific topic according to a keyword extraction method according to an embodiment of the present invention.

Referring to FIG. 10, a change in emotion of the user over time can be confirmed using the calculated average of the emotion index (Score). In the case of Apple, the most significant change in emotional index is from April 13 to April 14, and for Samsung from March 27 to 28, it can be extracted as emotional change interval.

Then, I collected TF-IDF by collecting tweets from April 13th to April 14th for Apple and from March 27th to March 28th for Samsung.

11 is a table showing top keywords extracted by applying the keyword extraction method according to an embodiment of the present invention. As shown in FIG. 11, top 10 top keywords with high frequency of occurrence are extracted through TF-IDF.

In the case of 'kindergartner' and 'crayons' among the extracted keywords at the time when the affinity for positive affection related to Apple increased, a school in Maine, USA, distributed the iPad to all kindergartners on April 13th. For Samsung, 'un32d6500' represents the newly released 32-inch 3D Ready LED HDTV from Samsung. These social events affect positive emotional changes. Among the 10 keywords, one keyword for Apple and two keywords for Samsung affect positive affects.

This shows that there is a great possibility in guessing social events that affect the emotional change of actual consumers when the association between the keyword extracted through the method proposed by the present invention and the social event is examined.

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

Claims (21)

In a method for extracting keywords that affect the emotional change of a public in a blog,
An emotion index calculating step of calculating an emotion index indicating information indicating emotion in a specific comment using the emotional vocabulary dictionary based on the vocabulary included in each comment in the blog;
A sensibility change section extracting step of detecting a change amount of the emotion index and extracting a sensibility change section of the blog; And
And a keyword extracting step of collecting comments of the emotional change section and extracting key keywords affecting emotional change in the emotional change section.
The method according to claim 1,
The method of claim 1, further comprising collecting at least one blog including a word for a specific topic at each reference time.
The method according to claim 1, wherein the emotion index calculating step
And extracting the refined data by removing a function word that does not serve to represent the contents of the sentence through the data preprocessing process for each comment. Way.
The method according to claim 1,
The emotional vocabulary dictionary is a vocabulary dictionary that quantifies the degree of a positive or negative semantic direction of a specific vocabulary. The keyword extraction method affects the emotional change of the public in a blog.
The method according to claim 1, wherein the emotion index calculating step
Calculating a positive index and a negative index for a vocabulary in each comment using the emotional vocabulary dictionary; And
And calculating a value obtained by subtracting the sum of the negative indices from the sum of the affirmative indices as an emotion index of each comment.
6. The method according to claim 5, wherein the emotion index calculating step
Calculating an emotion index of each of the comments existing within a reference time and calculating an interval emotion index; and extracting a keyword that affects the emotional change of the public in the blog.
7. The method according to claim 6, wherein the emotion index calculating step
And calculating the interval emotion index using an average of the emotion exponents of the respective comments existing within the reference time.
7. The method of claim 6, wherein the emotional change interval extraction step
And calculating a period in which the temporal change of the interval emotion index is greatest as the emotion change interval.
7. The method of claim 6, wherein the emotional change interval extraction step
And calculating a period of time in which the temporal change of the interval emotion index is greater than a predetermined threshold value as the emotion change interval.
2. The method according to claim 1,
(N is an arbitrary natural number) having a relatively high appearance frequency by using the TF-IDF (Temporary Frequency-Inverse Document Frequency) as the main keyword of the sensibility change section And extracting a keyword from the blog, the keyword being affected by a change in emotion of the public in a blog.
11. The method according to claim 10,
And extracting the main keyword based on an objective vocabulary excluding the vocabulary used in calculating the emotion index among the contents of the comment of the emotional change section. Way.
The method according to claim 1,
Wherein the blog is a micro-blog, wherein the blog is a micro-blog.
The method according to claim 1,
Wherein the comment includes at least one of a tweet, a reply, and a comment.
An apparatus for extracting keywords that affect a change in emotion of a public in a blog,
An emotion index calculating unit for calculating an emotion index indicating a degree of emotion in a specific comment by using an emotional vocabulary dictionary based on a vocabulary included in each comment in the blog;
A sensibility change section extracting section for detecting a change amount of the emotion index and extracting a sensibility change section of comments in the blog; And
And a keyword extracting unit for collecting comments of the emotion change period and extracting key keywords affecting the emotion change in the emotion change period.
15. The method of claim 14,
And a blog collection unit for collecting at least one blog including a word for a specific topic at each reference time.
15. The method of claim 14,
Wherein the emotional vocabulary dictionary is a vocabulary dictionary in which the degree of a positive or negative semantic direction of a specific vocabulary is quantified and expresses the degree of the positive or negative semantic direction of the specific vocabulary.
15. The apparatus of claim 14, wherein the emotion index calculating unit
A first calculation unit for calculating positive and negative indices for the vocabulary in each comment using the emotional vocabulary dictionary; And
And a second calculation unit for calculating a value obtained by subtracting the sum of the indefinite exponent from the sum of the affirmative indices as an emotion index of each comment.
15. The apparatus of claim 14, wherein the emotion index calculating unit
And calculating an interval emotion index using the average of the emotion indexes of the comments included in the reference time.
19. The method according to claim 18, wherein the emotional change section extracting section
Wherein a section in which the temporal change of the section emotion index is greatest is calculated as the emotion change section.
15. The apparatus of claim 14, wherein the keyword extracting unit
(N is an arbitrary natural number) having a relatively high appearance frequency by using the TF-IDF (Temporary Frequency-Inverse Document Frequency) as the main keyword of the sensibility change section Extracting a keyword that affects the emotional change of the public in a blog.
A system for extracting keywords that affect the emotional change of the public in a blog,
The data in the blog and the data included in the emotional vocabulary dictionary are received from the database and an emotional index indicating the degree of emotion in the specific comment based on the vocabulary included in each comment in the blog is calculated using the emotional vocabulary dictionary Extracting a sensibility change section of comments in the blog by detecting a change amount of the emotion index, collecting comments of the sensibility change section, and extracting a main keyword influencing emotion change in the sensibility change section, ;
A database for storing data related to the blog and the emotional vocabulary dictionary related data; And
And a user device for providing a comment on the blog, transmitting the key keyword extraction request to the keyword extracting device, and receiving the extracted main keyword from the keyword extracting device. A Keyword Extraction System Affecting Change.

Images