KR101122335B1 - Apparatus and method for analyzing information diffusion of topic based in the blog world - Google Patents

Abstract

Disclosed are an apparatus and method for analyzing topic spreading information in a blog world. The topic information spreading and analyzing device in the blog world includes a classification unit for classifying posts registered in a blog by a topic, and a preprocessing unit for associating a plurality of blogs in which the post is registered based on the spread history information for the post and forming a connection network. And, and using the network, the processing unit for analyzing the information spread for the posts classified by the topic.

Description

APPARATUS AND METHOD FOR ANALYZING INFORMATION DIFFUSION OF TOPIC BASED IN THE BLOG WORLD}

Embodiments of the present invention relate to an apparatus and method for analyzing information spread by topic in a blog world, which can analyze information spread by topic for a post registered in a blog and more accurately predict the spread probability between blogs.

A blog is a representative service means that allows users to express their ideas online in the form of posts. At this time, an online social network consisting of relationships between blogs is called a blog world. A blog may have various relationships with other blogs through service functions provided (eg, scraping, trackback, and commenting). As the ripple analysis information of the posts in the blog world is usefully used as basic data for the activation of the blog world, marketing to the blog users, research on the ripple analysis method in the blog world is being actively conducted.

The existing ripple analysis method in the blog world forms a network based on the ripple history information of all posts, assigns ripple probability between blogs, and then uses an independent propagation model to generate ripple analysis information about the posts in the blog world. Can be derived.

However, since the ripple analysis method does not consider the subject of the user or the post, the ripple analysis by topic is not easy. In addition, since the ripple analysis method does not consider a blog user that may have a ripple occurrence of a post, it is difficult to accurately predict the ripple in the blog world.

The present invention has been made in order to solve the above problems, based on the ripple history information for the post registered in the blog to analyze the spread of information on the topic by topic, in consideration of the potential candidate blogs, It aims to more accurately predict the spread probability between blogs.

In order to achieve the above object, the information spreading analysis apparatus for each topic in the blog world is to classify the posts registered in the blog by the topic, based on the spread history information on the posts and associates between the plurality of blogs registered the posts; It may be configured to include a pre-processing unit constituting a network, and a processing unit for analyzing the information spreading for the posts classified by the topic using the network.

In addition, as a technical method for achieving the above object, the method for analyzing the information spread by topic in the blog world is to classify the posts registered in the blog by the topic, the plurality of posts are registered based on the spread history information for the post Associating blogs and forming a network, and analyzing information spreading for posts classified by the topic using the network.

According to an embodiment of the present invention, based on the ripple history information for the posts registered in the blog to analyze the information ripples on the topic by topic, in consideration of the potential blog candidates more ripple, the ripple probability between blogs more accurately It can be predicted.

1 is a view showing the configuration of a topic-specific information spread analysis device in the blog world according to an embodiment of the present invention.
2 is a flowchart illustrating a method of analyzing information spread by topic in a blog world according to an embodiment of the present invention.
3 is a diagram showing a result of predicting the spread of a post about travel in a topic-specific information spread analysis device.
FIG. 4 is a diagram illustrating a result of predicting the spread of a post regarding English language in the information spreading analysis apparatus for each topic.
FIG. 5 is a diagram illustrating the accuracy of a connection network in which a ripple history is newly generated in a prediction interval, although a ripple history has not previously existed.
FIG. 6 is a diagram illustrating a phenomenon analysis result for five topics in a topic-specific information spreading and analyzing device.

Hereinafter, with reference to the accompanying drawings will be described in detail the apparatus and method for analyzing the information spread by topic in the blog world according to an embodiment of the present invention.

1 is a view showing the configuration of a topic-specific information spread analysis device in the blog world according to an embodiment of the present invention.

Referring to FIG. 1, an apparatus 101 for analyzing information per topic in a blog world according to an exemplary embodiment of the present invention includes a classification unit 103, a preprocessor 105, a processor 111, and a database 113. do.

The classification unit 103 classifies posts registered in a blog by a topic. For example, the classification unit 103 may classify posts registered in a blog based on a theme such as 'travel', 'English', or 'cooking'.

The preprocessor 105 associates a plurality of blogs on which posts are registered and forms a connection network based on the ripple history information on posts classified by subject. Here, the preprocessor 105 may configure the connection network using the propagation history information maintained in the database 113.

The preprocessing unit 105 may configure a connection network for a candidate blog that is determined to be registered with the first post based on the analyzed information spread when the connection network is configured. In this case, the preprocessor 105 may configure a blog as a node, and configure a connection network including trunk lines connected between the nodes. Thereafter, the preprocessor 105 may assign the propagation probability of the calculated post to the edge connected to the corresponding node.

  The preprocessing unit 105 configures a network in consideration of not only blogs in which posts have spread, but also candidate blogs that may have spread, thereby providing an environment in which the spread probability between blogs can be more accurately predicted.

In detail, the preprocessing unit 105 searches for a first blog that registers the first post and the second post, and a second blog that forms a connection network in connection with the second post, and searches for the searched second blog. If the first post is not registered, the second blog may be determined as a candidate blog and a connection network related to the first post may be configured between the first and second blogs. This distinguishes candidate blogs, considering that if a post on a particular topic spreads between blogs, posts on other topics may also spread.

For example, the preprocessing unit 105 searches for a first blog that registers a post about 'travel' and a post about 'English' and a second blog that forms a network in connection with a post about 'English', If a post about 'travel' is not registered in the second blog, the second blog may be determined as a candidate blog and a connection network related to the post about 'travel' may be configured between the first and second blogs. In other words, if the post about 'English' is spread between the first and second blogs, the preprocessing unit 105 may spread the post about 'Travel', so that the first network may be connected to the post about 'Travel'. Can be configured between the second blog.

In addition, the preprocessing unit 105 searches for a first blog for registering the first post and the second post, a third blog for registering a third post on the same topic as the first post, and the third post. The blog (or the first blog) may be determined as a candidate blog and a connection network associated with the first post (or the third blog) may be configured between the first and third blogs.

), 상기 연결망이 구성된 블로그로 파급되는 상기 제1 게시물과 다르게 분류된 제2 게시물의 비율에 관한 값(

), 상기 연결망이 구성되지 않는 블로그로 파급되는 제1 게시물의 비율에 관한 값(

), 또는 상기 연결망이 구성된 블로그에 등록된 상기 제1 게시물의 수에 관한 값(

) 중 어느 하나를 상기 인자값으로 정의할 수 있다.The preprocessor 105 may calculate the propagation probability of the post using the factor value for the configured network. Here, the preprocessing unit 105 is a value for the ratio of the first post spread to the blog composed of the network (

), A value relating to a percentage of second posts classified differently from the first post that the network is spread to

), A value relating to the percentage of first posts that spread to blogs that are not

), Or a value relating to the number of first posts registered to a blog configured with the network (

) May be defined as the factor value.

)은, 주제 K를 만족하는 블로그 A의 게시물 중에서 블로그 B에게 파급된 비율을 나타내는 인자값으로, 블로그 A가 주제 K에 대하여 블로그 B에게 미친 영향을 의미한다.First argument value (

) Is a factor value representing the proportion of blog A that spreads to blog B among the posts of blog A satisfying the topic K, and indicates the influence that blog A has on blog B on topic K.

을 산출할 수 있다. 구체적으로, 전처리부(105)는

(블로그 A가 가지고 있는 게시물의 집합)의 모든 게시물들의 파급 유효도를 계산하여, 이를 정규화한 값들의 총합을 계산한다. 상기 계산된 총합은 블로그 A가 주제 K에 대한 파급을 의도한 전체 규모라고 볼 수 있다. 전처리부(105)는

의 모든 게시물들의 파급 유효도를 정규화한 값들의 총합 중에서

에 포함된 모든 게시물들의 파급 유효도 값들을 정규화하여 합을 계산 함으로써, 실제로 블로그 B에게 파급된 비율을 산출하여, [수학식 1]과 같이 나타낼 수 있다.The preprocessor 105 uses a probability assigning technique using an effective score, which is a conventional spread probability assigning technique,

Can be calculated. Specifically, the preprocessor 105

Compute the ripple validity of all the posts (the set of posts that Blog A has) and calculate the sum of the normalized values. The calculated sum can be seen as the total size of blog A intended to spread to topic K. The preprocessor 105

Of the sum of the normalized ripple validity of all posts in

By calculating the sum by normalizing the ripple validity values of all the posts included in, the ratio actually spread to the blog B may be expressed as shown in [Equation 1].

는

에서 블로그 B에게 파급된 게시물들의 집합한다.Here, norm_score (Di) means a value obtained by normalizing the ripple validity value of Di,

Is

A collection of posts spread to Blog B.

)은, 주제 K가 아닌 블로그 A의 게시물 중에서 블로그 B에게 파급된 비율을 나타내는 인자값으로, 블로그 A가 K를 제외한 주제에 대하여 블로그 B에게 미친 영향이며, 다른 모든 주제들에 대하여 전반적으로 블로그 A 가 B에게 미친 영향을 의미한다고 볼 수 있다. Second argument value (

) Is a factor that indicates the percentage of blog A's posts that are spread to blog B, not topic K. This is the effect blog A has on blog B on topics other than K, and blog A on all other topics. Can be said to mean the effect on B.

와 마찬가지로, 유효 점수를 이용한 확률 부여 기법을 이용하여

을 산출하여, [수학식 2]과 같이 나타낼 수 있다.The preprocessor 105

Similarly, using the probability assigning technique using the effective score

It can be calculated as shown in [Equation 2].

)은, 블로그 A가 B를 제외한 블로그들에게 주제 K에 대하여 미치는 영향력을 나타내는 인자 값으로, 블로그 A가 B를 제외한 주제 K에 대한 이웃(neighbor)들에게 미친 영향을 의미한다.Third argument value (

) Is a factor indicating the influence of blog A on topic K on blogs except B, and indicates the effect blog A has on neighbors on topic K except B.

와 마찬가지로, 유효 점수를 이용한 확률 부여 기법을 이용하여

을 산출하여, [수학식 3]과 같이 나타낼 수 있다.The preprocessor 105

Similarly, using the probability assigning technique using the effective score

It can be calculated as shown in [Equation 3].

)은, 주제 K에 대한 블로그 B의 관심도를 나타내는 인자값으로, 블로그 B가 가지고 있는 주제 K를 만족하는 게시물의 수를 의미한다. 이때, 블로그 B의 주제 K에 대한 관심이 클수록, 이후에 주제 K에 대한 게시물을 파급할 가능성이 커지게 된다. The fourth argument value (

) Is a parameter value indicating the degree of interest of the blog B in the topic K, and means the number of posts satisfying the topic K of the blog B. At this time, the greater the interest in the topic K of the blog B, the greater the possibility of spreading posts on the topic K later.

As a configuration for calculating / predicting the propagation probability, the preprocessor 105 includes an analyzer 107 and a predictor 109.

The analysis unit 107 distinguishes a part of the network from the pre-connection network and another part of the network from the post-connection network, and checks the pre-factor value corresponding to the pre-connection network and the post-factor value corresponding to the post-connection network. A regression equation may be derived that includes the identified prefactor value as an independent variable and the propagation probability in the back network calculated as the dependent variable as a dependent variable. At this time, the analysis unit 107, for example, based on the set propagation time, can distinguish the front and back network from the connection network.

The prediction unit 109 may substitute the latter factor value into the regression equation to predict the propagation probability of the connection network configured for the candidate blog where the post is determined to be registered.

The processor 111 analyzes the spread of information on posts classified by topic using the connection network. That is, the processor 111 may apply the propagation model (eg, independent propagation model) to the connection network to which the propagation probability is assigned, and analyze the information propagation on the post classified by the subject.

The database 113 stores the ripple history information for the post. In this case, the database 113 may classify and store the ripple history information on the post by subject, and may also store the ripple history information by ripple time.

2 is a flowchart illustrating a method of analyzing information spread by topic in a blog world according to an embodiment of the present invention.

Referring to FIG. 2, the information spreading analysis apparatus classifies a post registered in a blog by a topic 201, and associates a plurality of blogs in which a post is registered based on the spread history information on posts classified by a topic. (203).

In this case, the information spreading analysis apparatus may configure a connection network for the candidate blog that is determined to be registered with the first post, based on the analyzed information spreading. The information ripple analysis apparatus can more accurately predict the spread probability between blogs by constructing a network in consideration of not only blogs where the posts have spread, but also candidate blogs that may have spread.

In detail, the information spreading analysis apparatus searches for a first blog for registering the first post and the second post, a second blog constituting a connection network in association with the second post, and the searched second blog. If the first post is not registered, the second blog may be determined as a candidate blog and a connection network related to the first post may be configured between the first and second blogs.

In addition, the information spreading analysis device searches a first blog for registering the first post and the second post, and a third blog for registering a third post on the same topic as the first post, and the third blog. The first blog may be determined as a candidate blog, and a connection network associated with the first post (or third blog) may be configured between the first and third blogs.

The information propagation analyzing apparatus calculates the propagation probability of the post using the factor value for the configured network (205).

), 상기 연결망이 구성된 블로그로 파급되는 상기 제1 게시물과 다르게 분류된 제2 게시물의 비율에 관한 값(

), 상기 연결망이 구성되지 않는 블로그로 파급되는 제1 게시물의 비율에 관한 값(

), 또는 상기 연결망이 구성된 블로그에 등록된 상기 제1 게시물의 수에 관한 값(

) 중 어느 하나를 상기 인자값으로 이용하여, 상기 게시물의 파급 확률을 산출할 수 있다.At this time, the information spreading analysis device is a value (ratio) regarding the ratio of the first post spreading to the blog that the network is configured (

), A value relating to a percentage of second posts classified differently from the first post that the network is spread to

), A value relating to the percentage of first posts that spread to blogs that are not

), Or a value relating to the number of first posts registered to a blog configured with the network (

) Can be used as the factor value to calculate the propagation probability of the post.

In detail, the apparatus for analyzing information spreads a part of the network into an all-connection network, and another part of the network into a back connection network, and the pre-factor value corresponding to the pre-connection network and the after-factor value corresponding to the after-connection network. In addition, a regression equation including the identified prefactor value as an independent variable and the propagation probability in the back network calculated using the postfactor value as a dependent variable may be derived. Subsequently, the information propagation analysis apparatus may estimate the propagation probability of the connection network configured for the candidate blog where the post is determined to be registered by substituting the latter factor value into the regression equation.

The information ripple analysis apparatus analyzes the information ripple for the post classified by topic using the connection network (207). In other words, the information ripple analysis apparatus may analyze the information ripple for posts classified by topic by applying a propagation model (for example, an independent propagation model) to the network to which the ripple probability is given.

According to an embodiment of the present invention, based on the ripple history information for the posts registered in the blog to analyze the information ripples on the topic by topic, in consideration of the potential blog candidates more ripple, the ripple probability between blogs more accurately It can be predicted.

Hereinafter, as a method of analyzing information spread by topic in the blog world, in order to verify the usefulness of the proposed prediction model, a result of performing predictive analysis to predict the spread of information to be generated later, and analysis of information spread occurring at the present time The following describes the results of the phenomenon analysis to determine the degree of accuracy.

 Experimental Environment

 Thematic information ripple analysis device in the blog world is a blog about five topics of 'travel', 'English', 'cuisine', 'car' and 'soccer' that many users (blogs) in the blog world spread information. Performance was analyzed in the network, and the number of blog users and the number of posts related to each topic are shown in [Table 1]. It is assumed that posts satisfy only one topic, and that the user may be interested in various topics.

For comparison of performance analysis, the topic-based blogging network using the probability-providing method using the existing effective score (Naive), the topical information spread analysis method (Topic) according to the present invention, the probability-providing method using the effective score (ES : Effective Score), technique CST1 (Constant 1%), which equally gives 1% probability to all edges, and technique CST5 (Constant 5%), which gives 5% probability equally to all edges.

In order to measure the performance of the above techniques, the apparatus for analyzing the information of each topic in the blog world is based on the propagation record generated by performing the independent propagation model on the records that are actually distributed in the blog world and the blog networks configured through each technique. Was compared. As a measure of comparison, F-measure calculated by using recall, precision and responsiveness and precision, which are widely used in the field of information retrieval. Responsiveness and precision are mutually exclusive. That is, the precision is low when the response has a very large value, and the response is low when the precision is very large. This harmonic mean value of responsiveness and precision is F-measure.

<Predictive Analysis>

)을 ag1, 두번째 인자값(

)을 ag2, 세번째 인자값(

)을 ag3, 네번째 인자값(

)을 ag4로 표기한다.The thematic information ripple analysis device in the blog world divided the data used in the analysis into three sections of the same size to measure the prediction accuracy of the thematic information ripple analysis technique. Subsequently, the thematic information spreading analysis apparatus in the blog world derives a formula for calculating the probability of spreading information using the factors obtained in the interval 1 and the probability of the distribution in the interval 2, and substitutes the values of the factors obtained in the interval 2 in the calculation formula. The probability of dissemination of information by topic was calculated. The thematic information ripple analysis device in the blog world compares the results of the independent propagation model on the blog network consisting of the probability of information ripple with the actual ripple history of section 3. Hereinafter, the four parameter values used as the independent variables in the regression analysis are described for convenience.

) Ag1, the second argument value (

) Ag2, the third argument value (

) To ag3, the fourth argument (

) Is written as ag4.

3 is a diagram showing a result of predicting the spread of a post about travel in a topic-specific information spread analysis device.

The regression analysis on the topic 'Travel' revealed that only ag2 was selected from the four factor values according to the step selection method, and the coefficient of determination (R squared) was 0.446. This means that 44.6% of the dependent variables are explained by independent variables. As a result of analysis of variance to verify whether the regression model is significant, the probability of significance is 0.000, indicating that the derived regression model is statistically significant. The results of verifying the statistical significance of the derived regression constants, regression coefficients, and regression coefficients can be represented by Table 2 as a regression model of the subject 'travel'.

The probability of the constant and the regression coefficient was 0.000, which showed statistically significant results. Regression equation F _traveling on the subject "travel" obtained through the regression analysis, is shown in [Equation 4].

The propagation probability of each topic is calculated using F _travel , and the independent propagation model is performed using this probability as shown in FIG. 3.

Referring to FIG. 3, the x axis represents comparison techniques and the y axis represents accuracy. First of all, the responsiveness of the technique is 0.31, which is about 2.4 times better than that of 0.13, which is 0.33. The precision of the technique is 0.68, which is about 91% of that of 0.75, which is 0.75. Looking at the F-measure, the technique Topic was 0.42, which showed about 1.8 times better performance than the 0.24 technique ES, and about 1.9 times the performance improvement compared to the 0.22 technique Naive, respectively, compared to the techniques CST1 and CST5. The performance improvement was about 42 times and about 11 times. Looking at the characteristics of each comparison technique, the method ES showed a result of little difference in responsiveness and precision. In the case of the technique Naive, the response is low while the precision is very high. Through this, the network consisting of the technique Naive resulted in less ripple performance due to the absence of edges when there is no ripple history, but the accuracy of the results is higher than that of other techniques. The technique Topic proposed in the present invention showed evenly high accuracy in response and precision, and showed the highest accuracy in F-measure, which is a harmonic mean value of response and precision. In the case of techniques CST1 and CST5, the accuracy is much lower than that of other techniques. Therefore, it is difficult to predict the spread of information by the method of giving the same probability to all edges.

FIG. 4 is a diagram illustrating a result of predicting the spread of a post regarding English language in the information spreading analysis apparatus for each topic.

According to the regression analysis on the subject 'English', only ag2 was selected from the four factors according to the step selection method, and the coefficient of determination (R-squared) was 0.403. This means that 40.3% of the dependent variables are explained by independent variables. As a result of variance analysis to verify whether the regression model is significant, the probability of significance is 0.000, indicating that the derived regression model is statistically significant. The results of verifying the statistical significance of the constants, regression coefficients, and regression coefficients of the derived regression model are regression models of the subject 'English', and can be represented by [Table 3].

The probability of the constant and the regression coefficient was 0.000, which showed statistically significant results. The regression equation F _English on the subject 'English' derived from the regression analysis is shown in [Equation 5].

The result of the information propagation probability of each subject is calculated using F _English , and the independent propagation model is performed using this probability as shown in FIG. 4.

Referring to FIG. 4, the x axis represents comparison techniques and the y axis represents accuracy. First of all, the response topologies showed a performance improvement of about 1.5 times compared to the technique naive of 0.30, which is 0.30, and the precision of the technique topic was 0.66, which is about 75% of that of the technique Naive of 0.88. Looking at the F-measure, the technique Topic is 0.41, which is about 1.6 times better than the 0.2 ES, which is about 1.2 times better than the 0.3 Nave, which is about 0.3 times, and compared to the techniques CST1 and CST5, respectively. The performance improvement was about 40 times and about 10 times. Looking at the characteristics of each comparison technique, the technique ES showed little difference in responsiveness and precision as in the topic 'Travel'. In the case of the technique Naive, the response was higher than the result of the subject 'travel', and there was almost no difference from the technique ES. This is related to the number of blog users involved in each topic. The number of users related to 'English' is larger than the topic 'Travel', and the larger the number of users related to the topic, the larger the network size of the blog network using the technique Naive. The larger the size of the blog network that spreads like this, the more results you get and the more responsive it is. The proposed Topic shows high accuracy in response and precision, and the highest accuracy in F-measure, which is the harmonic mean value of response and precision.

In addition, although the drawings for the theme 'cooking', 'car' and 'soccer' are not shown, the proposed technique Topic is better than other techniques in the prediction results for the subject 'cooking', 'car' and 'soccer'. High response and high accuracy in F-measure were found.

FIG. 5 is a diagram illustrating the accuracy of a connection network in which a ripple history is newly generated in a prediction interval, although a ripple history has not previously existed.

Referring to FIG. 5, when the average of the results of the five subjects is averaged, the technique Topic proposed by the F-measure is 0.37, which is about 1.5 times better than the technique ES of 0.25, and compared with the technique CST1 and the technique CST5, respectively. The performance improvement was about 37 times and about 9.3 times. In all subjects, the technique ES and the technique Topic appeared similar to the results of Experiment 1, but in the case of the technique Naive, if there was no previous history, no edges were established in the network, so no newly generated impacts could be predicted. .

FIG. 6 is a diagram illustrating a phenomenon analysis result for five topics in a topic-specific information spreading and analyzing device.

<Phenomena Analysis>

In order to analyze the spread of information that occurred at the present time, phenomenon analysis was performed on comparative techniques. In the phenomenon analysis, only one interval was used to derive the information propagation probability calculation for each subject of the proposed method.

Regression analysis on all subjects showed that the coefficient of determination (R-squared) was close to 0.999, meaning that most of the dependent variables were accounted for by independent variables.

Similarly, the variance analysis results verifying that the regression model is significant in all subjects, and the probability of significance is 0.000, indicating that the derived regression model is statistically significant. The significant probability of the constant and regression coefficients of the regression model was 0.000, which was statistically significant. The regression equations for each subject derived through the regression analysis are shown in [Table 4].

The regression equations for all subjects show that the regression coefficient of ag2 is very large. This means that the effect of user A on B is very large for all other subjects in the phenomenon analysis.

Referring to FIG. 6, the average of the results of the five subjects shows that the technique Topic proposed by the F-measure is 0.40, which is about 98% of the technique ES of 0.41, and about 1.5 times that of the technique Naive of 0.26. The performance of the system is improved by about 40 times and about 10 times as compared to the technique CST1 and the technique CST5. Both the results of the technique ES and the technique naive showed higher accuracy than the results of the predictive analysis, and because of the nature of the technique naive, only the users related to the subject constitute the network, so the more the number of users related to the subject, the higher the accuracy. As the proposed Topic is a model for predictive analysis, the phenomenon analysis showed similar or slightly lower accuracy than the technique ES.

As a result, the proposed Topic showed higher responsiveness, precision, and F-measure than other techniques in the predictive analysis, and it is found that it is useful for predicting new ripples that cannot be predicted by the technique Naive.

The subject information spread analysis method in the blog world according to an embodiment of the present invention may be implemented in the form of program instructions that can be executed by various computer means and recorded in a computer readable medium. The computer readable medium may include program instructions, data files, data structures, etc. alone or in combination. Program instructions recorded on the media may be those specially designed and constructed for the purposes of the present invention, or they may be of the kind well-known and available to those having skill in the computer software arts. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tape, optical media such as CD-ROMs, DVDs, and magnetic disks, such as floppy disks. Magneto-optical media, and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware device described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

As described above, the present invention has been described by way of limited embodiments and drawings, but the present invention is not limited to the above embodiments, and those skilled in the art to which the present invention pertains various modifications and variations from such descriptions. This is possible.

Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined not only by the claims below but also by the equivalents of the claims.

101: Apparatus for analyzing and spreading thematic information in the blog world
103: classification
105: preprocessing unit
111: processing unit
113: database

Claims (14)

Search for a second blog constituting a network in connection with a second post registered in a first blog, and among the searched second blogs, a blog that does not register the first post registered in the first blog as a candidate blog. A preprocessor configured to determine and form a connection network between the first blog and the candidate blog, the connection network associated with the first post; And
Prediction unit for predicting the propagation probability of the network configured for the candidate blog
Apparatus for analyzing the topic information in the blog world comprising a. The method of claim 1,
A classification unit classifying posts registered in a blog by a topic; And
A processing unit for analyzing the information spread for the posts classified by the topic using the network
Further comprising:
The preprocessing unit,
The subject-specific information spreading analysis device in the blog world, further comprising the connection network for the candidate blog in consideration of the analyzed information spreading. delete delete The method of claim 1,
The preprocessing unit,
The propagation probability of the post with respect to the configured network, the value of the ratio of the first post to the blog is configured to the network, the second post classified differently from the first post to the blog is configured Calculating based on any one of a value relating to a ratio, a value relating to a ratio of first posts propagated to a blog which is not configured, or a value relating to the number of first posts registered to a blog configured as the connecting network. , Thematic information ripple analysis device in the blog world. The method of claim 1,
The preprocessing unit,
Deriving a regression equation that distinguishes the network from the front network and the back network based on the set propagation time point, and includes a value associated with the front network as an independent variable and a propagation probability of the back network as a dependent variable. Analysis department
Apparatus for analyzing thematic information spread in the blog world, including. delete Searching for a second blog constituting a connection network in association with a second post registered in the first blog;
Determining, from among the searched second blogs, a blog which does not register a first post registered in the first blog as a candidate blog;
Configuring a connection network associated with the first post between the first blog and the candidate blog; And
Estimating the propagation probability of the network configured for the candidate blog
Thematic information spread analysis method in the blog world comprising a. The method of claim 8,
Categorizing the posts registered in the blog by the topic; And
Analyzing the information spread for the posts classified by the subject using the network;
Further comprising:
The step of configuring the network,
Configuring the connection network for the candidate blog further considering the analyzed information spread
Thematic information spread analysis method in the blog world comprising a. delete delete The method of claim 8,
Predicting the propagation probability of the network,
The propagation probability of the post with respect to the configured network, the value of the ratio of the first post to the blog is configured to the network, the second post classified differently from the first post to the blog is configured Calculating based on any one of a value relating to a ratio, a value relating to a ratio of first posts propagated to a blog which is not configured, or a value relating to the number of first posts registered to a blog configured as the connecting network. step
Thematic information spread analysis method in the blog world comprising a. The method of claim 8,
Deriving a regression equation that distinguishes the network from the front network and the back network based on the set propagation time point, and includes a value associated with the front network as an independent variable and a propagation probability of the back network as a dependent variable. step
Thematic information spread analysis method in the blog world further comprising. delete

Images