KR20100115600A - Method and apparatus for online community post searching based on interactions between online community user and computer readable recording medium storing program thereof - Google Patents

Abstract

An online community post search technology is disclosed that provides specialized results using the proficiency or popularity scores of online community users defined by quantifying the interactions among online community users such as comments, recommendations, and links, and the semantic similarity score between search terms and online community posts. do. According to an aspect of the present invention, first information about a plurality of online community users and second information about an online community post written by the online community user are received, and the online community post written by a specific online community user among the online community users is input. Obtain third information about the interaction formed by the online community user different from the specific online community user among the online community users with respect to each of the online community users; 3 information is used to calculate a first score for each of the online community users, the first score of the specialty of the specific online community user, and the semantic similarity between the input search word and the online community post Calculating a second score on, and provides the first score and the second score by using the online community user or the online community user interaction based on an online community post-search method for searching the online community post. According to the present invention, it is possible to provide a search result reflecting various intentions of an online community post search user even with a simple search word.

Description

Method and Apparatus for Online Community Post Searching Based on Interactions between Online Community User and Computer Readable Recording Medium Storing Program

The present invention relates to an online community post retrieval method, apparatus, and a computer readable recording medium having recorded thereon a program for executing the method. Specifically, the present invention uses the expertise score or popularity score of the online community user defined by quantifying the interaction between online community users such as comments, recommendations, and links, and the meaning similarity score between the search word and the online community post to An online community post retrieval technique based on interaction between online community users providing specialized results.

Recently, online community posts, which are represented in various forms on-line, such as blogs, cafes, bulletin boards, and the like, are used for social media. It is rapidly spreading as a media. Online community websites on specific hobbies or academic topics such as photography, music, science, and digital equipment are popular. Some of these online community Web sites have tens of thousands of registered users and millions of user-created content (UCC).

Blogs are one of the representative forms of these online community posts, and statistics show that the number of blog users in Korea ranges from millions to 10 million. In the lives of modern people living with the Internet, blogs touch all aspects of human history such as society, economy, culture, and politics, and their influence is also increasing. The cyberspace shared by blog users, or bloggers, is called a blogosphere. The blog space has something in common with the general web in that it maintains a connection with the other party through a hyperlink. However, blogs differ from the general web in that they provide a personally oriented space. Individual blogs express their interests such as politics, economy, culture, art, sports, and hobbies freely through a post called post, which is composed of posts and pictures.

Blogs have a more formal form than regular web pages, and they also focus on personal concerns in terms of content. This opens up the possibility of providing search engines with more sophisticated and powerful search features just for blogs. In fact, many recent technical developments on blog search have been made at home and abroad, and many research papers have been published.

Currently, domestic and overseas blog sites provide search results based on three types of information. First, there is a search ranking function based on simple statistics of blog posts. Examples include comment order, track back best, attention degree, consensus, recommendation, popular article, star rating, eye-catching blog, and topical bloggers. Second, there is a search function that determines the ranking based on the time the blog post is created. Examples include new posts, Today's Hot Blogs, Today's Hot Posts, Most Popular, Real-Time Popular Posts, Real-Time New Posts, Newest, and Top Top Bloggers of the Week. Third, there is a search function in which a search rank is determined based on linguistic matching (TF * IDF) between a query keyword such as a tag, a keyword, and an accuracy order, and a keyword of a blog post. In short, the prior art search ranking has been based on three factors: simple statistics, creation time, and linguistic matching.

On the other hand, research papers published in relation to blog search can be classified into two main topics: influential blogger discovery technology and blog ranking technology. Briefly introducing the conventional research for each of these are as follows.

First, there is research on techniques for finding notable bloggers. Nitin Agarwal (2008) et al. Address the issue of finding influential bloggers in blog spaces. This paper proposes four statistical features for measuring the social gestures of bloggers. The number of citations as a recognition feature, the number of comments as an activity generation feature, the number of other blog citations as an originality feature, and the length of a post as an eloquence feature. In order to calculate the influence score of the blog post, the weighted sum of each feature value is calculated, and each blogger calculates the highest value of his blog post score as his or her recognition score. Bi Chen et al. (2007) proposed a method for predicting future topics by analyzing the behavior patterns of bloggers on the temporal and social levels. Xiaodan Song et al. (2007) proposed a technique to find an influential opinion leader in the blog space by introducing the concept of influence rank. Craig Macdonald et al. (2008) deals with the problem of dividing a unit of search into an entire blog or a blog post to find an expert on a particular topic of interest. In a paper by Akshay Java et al. (2006), an influence model was proposed to find an influential blogger. Meanwhile, P. Jurczyk et al. (2007) propose a problem of finding an authoritative user in a question / answering type system such as Naver knowledge search.

Second, there is a study on the technology of ranking blogs. A paper by A. Kritikopoulos et al. (2006) proposed a blog ranking technique using implicit hyperlinks. For example, suppose two blog posts with the same topic are linked by implicit hyperlinks. These implicit hyperlinks can compensate for sparse link relationships between blog posts. In addition to the same topic, implicit hyperlinks have also been applied to bloggers and commenters of the same topic. The paper also proposed a blog ranking algorithm using these coefficients in the PageRank style disclosed in the paper by L. Page et al. (2008). Xiaochuan Ni et al. (2007) proposed a search technique for blog posts that contain useful or influential information. Kritsada Sriphaew et al. (2008) proposed a technique for searching cool blogs in consideration of the characteristics related to blog topics. For example, an important blog assumes that it has a clear topic, has a sufficient number of posts, and that topic consistency exists, and has presented a probabilistic model.

Taken together, we can see that existing bloggers and blog post search techniques are focused on one aspect of value: influence or intelligence. The most influential or informative bloggers or blog posts are focused on solving problems about something. The problems of the conventional blog post and blogger search technology have been recognized as problems in other forms of online community post and user search technology as well.

However, the search intentions, trends, and trends of users searching online community posts such as blogs vary greatly. Just because multiple online community post search users enter the same search term does not necessarily mean that their search purpose is the same. In this respect, it is difficult to say that the prior art, which only applies ranking based on one measure, has been able to meet various search needs of online community post search users.

An object of the present invention is to solve the above problems, and to develop an online community post search technology that can accommodate various intentions of search users using social network analysis and data mining technology. do.

In addition, it aims to provide a search result reflecting various intentions of the user with a simple search word by introducing a trend concept such as expertise and popularity.

In order to achieve the above object, a first aspect of the present invention includes receiving first information about a plurality of online community users and second information about an online community post written by the online community user, and among the online community users, Obtaining, according to an online community post written by a community user, third information about an interaction formed by an online community user different from the specific online community user among the online community users, for each of the online community users; The first score of the specialty of the specific online community user is calculated for each of the online community users by using the score of the professionalism and the third information, and the inputted search word and the online A method for searching online community posts based on interactions between online community users, which calculates a second score regarding semantic similarity between community posts, and searches for the online community user or the online community post using the first score and the second score. To provide.

In addition, the third information is the number of interactions, and the calculating of the first score comprises: performing a first operation comprising a multiplication of the third information and a score relating to the expertise of the different online community users; An online community post retrieval method based on interaction between online community users performing a second operation, which is performed for each online community user, and includes a sum of the results of the first operation for the different online community users.

In addition, the process of calculating the first score is the following equation

는 상기 Au에 속하는 온라인 커뮤니티 포스트에 대하여 상기 온라인 커뮤니티 사용자 중 상기 상이한 온라인 커뮤니티 사용자 v가 형성한 상호작용의 횟수,

는 상기 상이한 온라인 커뮤니티 사용자 v가 형성한 모든 상호작용의 횟수, d는 최소 영향력을 나타내는 감쇄 인자(damping factor)인 온라인 커뮤니티 사용자 간 상호작용 기반 온라인 커뮤니티 포스트 검색 방법을 제공한다. ER (u) is the first score of the particular online community user u, ER (v) is the score relating to the expertise of the different online community user v, Au is the specific online community user u Of online community posts written by

Is the number of interactions formed by the different online community users v among the online community users for the online community posts belonging to the Au ,

Is a number of all interactions formed by the different online community users v, and d provides a method for searching online community posts based on interactions between online community users, which is a damping factor indicating a minimum impact.

Also, the online community users are sorted in order of the first score, and the third score about the popularity of the particular online community user is recalled by using a mapping function having a higher value as the sorted ranking is closer to the middle rank. Computation is performed for each online community user, and the search provides an online community post search method based on interaction between at least one of the first score or the third score and the online community user using the second score.

In addition, the mapping function provides an online community post search method based on interaction between online community users including a Gaussian distribution function.

In addition, the Gaussian distribution function is

Where PR (u) is the third score of the particular online community user u, x is the ordered rank of the particular online community user u, σ is the standard deviation of the ordered rank, μ is the Provides an online community post retrieval method based on interaction between online community users, which is the median of sorted rankings.

The present invention also provides an online community post search method based on interaction between online community users who receive a selection of a score to be used for the search among the first score or the third score from a user before inputting a search word.

In addition, the online community posts are divided into a plurality of categories, and among the online community users, each of which calculates a score on the expertise of the different online community users and a first score on the expertise of the specific online community user separately according to the category. Provides an interactive, online community post retrieval method.

In addition, the interaction provides an online community post search method based on interaction between online community users including any one or more of comments, comments, empathy, recommendation, trackbacks, scraps, and links.

In order to achieve the above object, a second aspect of the present invention provides a computer-readable recording medium having recorded thereon a program for executing any one of the above methods.

In order to achieve the above object, a third aspect of the present invention provides an online community user information storage unit for storing first information about a plurality of online community users and a first score regarding the expertise of the online community user, and the online community. Second information about an online community post created by a user and an interaction formed by an online community user different from the specific online community user among the online community users with respect to an online community post written by a specific online community user among the online community users An online community post storage that stores third information, a score about the expertise of the different online community users, and a first score about the expertise of the particular online community user using the third information. Using a professional computing unit for each of the online community users, a semantic similarity calculating unit for calculating a second score on semantic similarity between the input search word and the online community post, and the first score and the second score. Provided is an online community post search apparatus based on interaction between an online community user or an online community user including a search unit for searching for the online community post.

In addition, the third information is the number of interactions, and the professional operation unit is configured to perform a first operation on each of the different online community users, the first operation including a multiplication of the third information and a score about the expertise of the different online community users. And an adder for performing a second operation including a multiplier for performing a second operation and a summation of the results of the first operation for the different online community users.

In addition, the online community users are sorted in the order of the first scores, and a function having a higher value is mapped as the sorted rank is closer to a middle rank, thereby obtaining a third score on the popularity of the specific online community user. And a popularity calculation unit for calculating each user, wherein the search unit provides an online community post search apparatus based on interaction between one or more of the first score or the third score and an online community user using the second score. .

According to the present invention, it is possible to provide a search result that reflects various intentions of an online community post search user even with a simple search word.

In addition, by defining trends of expertise and popularity based on interactions between online community users and using them in search, it is possible to reflect changes in the online community post environment in real time.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description of the present invention, if it is determined that a detailed description of a known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted. The following terms are defined in consideration of the functions of the present invention, and may be changed according to the intentions or customs of the user, the operator, and the like. Therefore, the definition should be made based on the contents throughout the specification.

The present invention introduces the concept of trends as a measure of the user's various search intents, trends, trends, and the like. The contents of this trend-based search technology will be described in detail through embodiments of expertise and popularity.

For example, a user searching for a blog related to a movie does not search for a movie by one measure. Searches like "search for the most important movies" are meaningless. However, let's look at an example that reflects trends such as popularity and professionalism in movie search. Even if a search user searches for a blog related to the same genre of movies in the film field, some search users prefer the most popular films, and some search users will prefer movies with high professionalism or artistry. Trends or preferences of these search users are defined as trends. Thus, when the concept of trends is introduced into search technology, unlike the conventional technique of simply searching for a blog post that matches a movie title or actor's name, "search for a work with high professionalism" or "search for work with a mass" Can handle search requests. Searching blogs based on these trends cannot be done by simply comparing the statistics of blog posts such as views and referrals.

1 illustrates the interactions that a plurality of online community users form in each other's online community posts. Here we see the interaction relationships that three online community users, online community users A, B, and C, perform on each other's online community posts.

The online community post 110 of the online community user A posts three posts: a post A1 111, a post A2 112, and a post A3 113. The post A1 (111) has a comment 171 left by the online community user B, and the post A2 112 has a sympathy (172) message left by the online community user B and the online community user C has written in the post C1 (151). Link 174 is linked, and there is a recommendation 175 left by online community user C in post A3 113. The online community post 130 of the online community user B posts two online community posts, a post B1 131 and a post B2 132. Post B1 131 has a comment 173 left by online community user A, and post B2 132 can see that online community user C scraped 172 to post C2 152. The online community post 150 of the online community user C posts two online community posts, the post C1 151 and the post C2 152. Post C1 151 has a link 174 linking Post A2 112 of online community user A, and post C2 152, which is a post that scraped 177 post B2 132 of online community user B. It can be seen that is connected to the trackback 176 and the post A3 113 of the online community user A. The comment (171), empathy (172), comment (173), link (174), recommendation (175), trackback (176), scrap (177), etc. are different names and forms depending on the online community services provided. It is the interaction between multiple online community users and is common in that it is a quantifiable factor. In addition to the examples illustrated by such online community users, there may be various forms.

2 is a flowchart illustrating each step of an embodiment of an online community post search method based on interaction between online community users according to the present invention.

First, first information about a plurality of online community users and second information about an online community post written by the online community user are received or stored (S210). And to each of the online community users, third information about an interaction formed by an online community user different from the specific online community user among the online community users with respect to an online community post written by a specific online community user among the online community users. Acquire or store (S220). In operation S230, a first score regarding the expertise of the specific online community user is calculated for each of the online community users using the scores regarding the expertise of the different online community users and the third information. In operation S240, a second score regarding semantic similarity between the input search word and the online community post is calculated. Finally, the online community user or the online community post is searched using the first score and the second score (S250).

According to an embodiment, an online community user clustering modeling a social network inherent between users in an online community space in a graph form and classifying highly related online community users as sub communities. Can be performed. The subcommunity that appears as a result of such online community user clustering is composed of online community users that are representative of the field. The trend ranking value of the online community user is calculated for each online community user in the subcommunity. Trend ranking is a kind of score on expertise, popularity, etc., which can be achieved through social network analysis and data mining techniques. When a user query is given in the form of a search term, the online community post is finally obtained by combining the semantic similarity score between the user query and the online community post and the trend score value of the online community post. Ranking results will be returned. Trend scores of online community posts are obtained through trend ranking values of online community users that interact with or relate to the post. The same technology can be applied to all types of online communities, including blogs.

Hereinafter, an embodiment of the present invention will be described using the concept of social network analysis. Online community posts generated by online community users in the fields of photography, music, movies and the like can be ranked in two opposing perspectives: popularity and expertise. Some searchers will want to search for popular photos that are popular with the general public, while others will prefer photos that are highly professional, such as work from an expert point of view. As described above, the search user's tendency or preference may be defined as a trend.

The basic ideas of online community user rankings for these professional trends are as follows. If an online community post contains high-level or expert-level content, the online community post will get more response from experts than non-experts. Conversely, if a post is low to novice content, expert users will not respond to it.

, 그리고 온라인 커뮤니티 사용자 u가 쓴 Au에 속한 포스트들에 대하여 온라인 커뮤니티 사용자 v에 의해서 쓰여진 커멘트 집합을

라 할 때, 온라인 커뮤니티 사용자 u의 전문성 랭크값 ER(u)는 아래 수학식 1과 같이 공식화할 수 있다.From this inference, we can define an Expert Rank (ER), which is the expertise score of an online community user based on the interactions or relationships between the online community users. The ER value of any online community user u is influenced by the ER value of all other online community users v who have responded or interacted with comments, comments, empathy, recommendations, trackbacks, scraps, links, etc. to the posts created by u. Will receive. This section uses comments from various forms of interaction as an example. Set of posts written by the online community user u, | Au |

And a set of comments written by the online community user v for posts belonging to Au written by the online community user u.

In this case, the professional rank value ER (u) of the online community user u may be formulated as in Equation 1 below.

는 Au에 속하는 온라인 커뮤니티 포스트에 대하여 다른 온라인 커뮤니티 사용자 v가 형성한 상호작용의 횟수,

는 상이한 온라인 커뮤니티 사용자 v가 형성한 모든 상호작용의 횟수, d는 최소 영향력을 나타내는 감쇄 인자(damping factor).Where ER (u) is the professional score of a specific online community user u, ER (v) is the professional score of another online community user v, Au is a set of online community posts written by a particular online community user u,

Is the number of interactions formed by other online community users v for online community posts belonging to Au ,

Is the number of all interactions formed by different online community users v, d is a damping factor representing the least impact.

Equation 1 is only an embodiment of the present invention, and various modifications of Equation 1 are possible in calculating the professional score. In the present embodiment, the professional scores of other online community users are recursively used to calculate the professional scores of specific online community users. However, various other methods using interactions between online community users may be used. The calculation of the professional score may add consideration to the activity, recognition, sociality, etc. of online community users.

3 illustrates a result of arranging the professional scores according to the professional rankings of online community users. This corresponds to the ER distribution of the entire online community user derived by Equation (1). Each of the horizontal and vertical axes is plotted on a log scale. As the number of the online community user professional ranking or the online community user rank increases, that is, the professional rank decreases, the professional score or the ER value decreases. In the online community user professional rankings of the top 10, the professional score is about 1000 points or more, but when the online community user professional rankings of 1000 or more research results show that the professional score falls below 10 points.

4 illustrates a process of calculating a popularity score by mapping a Gaussian distribution function to a result of sorting the expertise score according to the expertise ranking of the online community user. First, the concept of popularity trend will be explained, and then the mapping process will be described.

The popularity of online community users is a concept that is closely related to professionalism, but not the opposite. This is because there may be many online community users with high professionalism and popularity, or users of online communities with low expertise and popularity. Popularity can be determined to be an online community user preferred by a high-level or expert user or a general online community user with a moderate level of expertise than a low-level or beginner user.

In this regard, the popularity score of the online community user may be defined through the ER value, which is the above-described expertise score. This can be achieved by sorting the online community users by ER value, mapping each online community user to a Gaussian distribution function, and applying weights. When using this method, the popularity score PR (Popularity Rank) of the online community user may be calculated by the following formula (2).

Where PR (u) is the popularity score of a particular online community user u, x is the rank sorted by professionalism of the specific online community user u, σ is the standard deviation of the ranks sorted by professionalism, μ is the rank Median of.

Equation 2 is merely an embodiment of the present invention, and various modifications of Equation 2 are possible in calculating the popularity score. In the present embodiment, the popularity is calculated through the mapping of the Gaussian distribution function from the ranks sorted in order of the expertise scores of specific online community users. May be

Referring to FIG. 4, it can be seen that a Gaussian distribution function 420 is mapped to a result 410 of arranging the professional scores according to the expertise rankings of the online community users. The Gaussian distribution function 420 may be mapped to have a maximum value near the median of the professional score alignment result 410. This mapping gives the online community users with moderate expertise the highest level of popularity.

In Figure 4 both the horizontal and vertical axes are shown on a logarithmic scale. Therefore, the median value of the professional score alignment result (410) is substantially the median of the log scale conversion value. If the total number of online community users is 10,000 (log 10,000 = 4), the expert score is ranked 100 (log 100 = 2). The degree of popularity score is given to the degree of online community users. For example, in FIG. 4, an online community user having an expert rank of about 100 ranks has an expert score 440 of about 100 points. The popularity score of 445 is calculated to be very high, around 3,000 points. In addition, although the expertise score 430 of an online community user having a professional rank of about 10 ranks is about 1,000 points, the popularity of the online community user, which is converted when mapping the professional rank of about 10 ranks to a Gaussian distribution function 420, is converted. It can be seen that the score 435 is calculated very low, around 10 points.

However, in some embodiments, the mapping function and the mapping scale may be applied in various forms. For example, if you draw a graph on the horizontal or vertical axis instead of the log scale, and map the Gaussian distribution function as it is, an online community of 5,000 or more based on a professional score of 10,000 users. The user can be given a maximum level of popularity score. The degree of popularity score given to the users of the professional rank can be determined and applied differently according to various environments in the online community.

Meanwhile, as a technique used in the search process, the following Equation 3 may be proposed as one of a method of calculating an online community post score based on a specific trend.

According to Equation 3, a query that is a user's query is finally obtained through a combination of a semantic similarity score between a user query using TF * IDF and the like and an online community post, and a score score of the trend score of the online community post. Will rank the online community posts and return the results. Equation 3 uses a kind of weighted average concept, but other types of equations that simultaneously reflect the semantic similarity score and the trend score value may be used.

According to an embodiment, the user may select and input in advance which of the various trends to perform a search based on various trends before the user inputs the initial search word. In this case, a plurality of trend-based online community post score calculation formulas corresponding to Equation 3 may be made for each selectable trend, and may be implemented to use corresponding calculation formulas according to a user's selection. In addition, when one semantic similarity score and a plurality of trend score values are combined to be applied to calculate an online community post score, it is possible to search by applying one or more trends simultaneously.

에 의해서 아래 수학식 4와 같이 계산될 수 있다.Online community post p's score for a particular trend is a function

It can be calculated by Equation 4 below.

Where Up is a set of online community users with a connection to p, and Pp is a set of posts with a connection to p.

는 트렌드에 따라서 다르게 설계될 수 있다. 본 발명의 일 실시예로서 설명된 전문성 및 대중성 이외의 다른 주요 트렌드의 경우 함수

의 설계에는 다양한 데이터 마이닝 기술이 접목될 수 있다. 적용 가능한 데이터 마이닝 기술의 예로는 1-mode/2-mode 사회적 연결망 분석, Markov chain random walk 모델 분석, 연관 규칙 탐색(association rule mining), 분류화(classification) 등이 있다.That is, the score of an online community post for a particular trend can be defined by the interaction with the online community user and the post that is connected to the post. Where function

Can be designed differently depending on the trend. Functions in the case of major trends other than professionalism and popularity described as an embodiment of the invention

Various data mining techniques can be combined in the design of. Examples of applicable data mining techniques include 1-mode / 2-mode social network analysis, Markov chain random walk model analysis, association rule mining, and classification.

Meanwhile, according to an embodiment, in order to efficiently apply trends such as professionalism or popularity, the online community may be divided into a plurality of categories, and information on professionalism may be separately calculated according to the categories. It is possible to produce a more accurate result than applying a global trend rank value, that is, a collective trend score, without distinguishing the category of each online community user and post. For example, if a post about a landscape photograph is evaluated by landscape photographers, it may be more accurate than if it is evaluated by model photographers. To this end, each online community user and post can be clustered or categorized by field and can be implemented to reflect a high percentage of trend scores belonging to the same field.

5 is a block diagram illustrating a schematic configuration of an embodiment of an online community post search apparatus based on interaction between online community users according to the present invention. According to the present embodiment, an online community user information storage unit 510, an online community post storage unit 520, a professionality calculator 530, a semantic similarity calculator 540, a searcher 550, and a popularity calculator 560 are configured. It can contain elements.

The online community user information storage unit 510 stores first information about the plurality of online community users and a first score about the expertise of the online community users. In addition, various information about the online community user may be stored in the online community user information storage unit 510. The online community user information storage unit 510 may transfer the expertise information 591 of the online community user, which is the first score, and the first information to the expertise calculation unit 530.

The online community post storage unit 520 relates to second information about an online community post written by an online community user and an interaction formed by a different online community user and a different online community user with respect to an online community post written by a specific online community user. Store the third information. In addition, various information about the online community post may be stored in the online community post storage 520. The online community post storage 520 may transmit the interaction information 592 and the second information, which are the third information, to the professionality calculator 530.

The expertise calculator 530 calculates, for each of the online community users, a first score about the expertise of a specific online community user by using scores and third information about the expertise of different online community users. Expertise operation unit 530 is a multiplier that performs a first operation for each of the different online community users and a first operation for different online community users, the first operation including multiplication of the third information and the score on the expertise of different online community users. It can be implemented to include an adder that performs a second operation that includes the summation of the results. The professionality calculator 530 may transmit the professional score 593, which is the first score, to the searcher 550 and the popularity calculator 560.

The semantic similarity calculator 540 calculates a second score regarding semantic similarity between the input search word and the online community post. The semantic similarity calculator 540 may transmit a semantic similarity score 595, which is a second score, to the searcher 550.

The search unit 550 searches for an online community user or an online community post using the first score or the third score, and the second score. The search unit 550 may provide a search result to a user by using a professional score 593 which is a first score, a semantic similarity score 595 which is a second score, and a popularity score 596 which is a third score.

According to an embodiment, the popularity calculator 560 may be included. The popularity computing unit 560 sorts the online community users in the order of the first scores, maps a function having a higher value as the sorted ranking is closer to the middle rank, thereby obtaining a third score on the popularity of a specific online community user. Calculate for each. In this case, the search unit 550 may be implemented to use any one or more of the first score or the third score and the second score. The popularity calculator 560 may transmit a popularity score 596, which is a third score, to the searcher 550.

The technical concept of the present invention is an online community that can share a post, which is a post written by a plurality of users, such as a blog, a cafe, a bulletin board, and the like. Applicable to an online community posting / post, the present invention is not limited to the specific embodiments mentioned in the embodiments of the present invention.

Modules, functional blocks or means of the present embodiment may be implemented in a variety of known elements, such as electronic circuits, integrated circuits, ASICs (Application Specific Integrated Circuit), each may be implemented separately, or two or more may be integrated into one Can be.

Although the embodiments have been described for the understanding of the present invention as described above, it will be understood by those skilled in the art, the present invention is not limited to the specific embodiments described herein, but variously without departing from the scope of the present invention. May be modified, changed and replaced. For example, the technique of the present invention may be applied to a picture, an image, etc., which may be displayed by a display such as an LCD instead of a character. Therefore, it is intended that the present invention cover all modifications and variations that fall within the true spirit and scope of the present invention.

1 illustrates the interactions that a plurality of online community users form in each other's online community posts.

2 is a flowchart illustrating each step of an embodiment of an online community post search method based on interaction between online community users according to the present invention.

3 illustrates a result of arranging the professional scores according to the professional rankings of online community users.

4 illustrates a process of calculating a popularity score by mapping a Gaussian distribution function to a result of sorting the expertise score according to the expertise ranking of the online community user.

5 is a block diagram illustrating a schematic configuration of an embodiment of an online community post-search apparatus based on interaction between online community users according to the present invention.

Claims (13)

Receiving first information about a plurality of online community users and second information about an online community post written by the online community user, Obtaining, for each of the online community users, third information about an interaction formed by an online community user different from the specific online community user among the online community users with respect to an online community post written by a specific online community user among the online community users. and, Calculate a first score for the specialty of the particular online community user for each of the online community users, using the scores for the expertise of the different online community users and the third information, Calculating a second score on semantic similarity between the received search word and the online community post, Searching for the online community user or the online community post using the first score and the second score How to search online community posts based on interactions between online community users. The method of claim 1, The third information is the number of interactions, The calculating of the first score comprises performing a first operation for each of the different online community users, the first operation comprising multiplication of the third information and a score about the expertise of the different online community users, wherein the different online community users To perform a second operation including a summation of the results of the first operation on How to search online community posts based on interactions between online community users. 3. The method of claim 2, The process of calculating the first score is represented by the following equation

Is based on where ER (u) is the first score of the particular online community user u, ER (v) is a score on said expertise of said different online community user v, Au is a collection of online community posts written by the specific online community user u,

Is the number of interactions formed by the different online community users v among the online community users for the online community posts belonging to the Au ,

Is the number of all interactions formed by the different online community user v, d is a damping factor representing the least impact How to search online community posts based on interactions between online community users. The method of claim 1, Sort the online community users in order of the first score, Calculating, for each of the online community users, a third score on the popularity of the particular online community user using a mapping function having a higher value as the sorted ranking is closer to the middle rank, At least one of the first score or the third score and the second score are used for the search. How to search online community posts based on interactions between online community users. The method of claim 4, wherein The mapping function includes a Gaussian distribution function. How to search online community posts based on interactions between online community users. The method of claim 5, The Gaussian distribution function is

Expressed as PR (u) is the third score of the particular online community user u, x is the sorted ranking of the specific online community user u, σ is the standard deviation of the sorted rank, μ is the median of the sorted rank How to search online community posts based on interactions between online community users. The method of claim 4, wherein Receiving a selection of a score to be used for the search among the first score or the third score from a user before inputting a search word How to search online community posts based on interactions between online community users. The method of claim 1, Divide the online community post into a plurality of categories, Calculating a score on the expertise of the different online community users and a first score on the expertise of the particular online community user separately according to the category. How to search online community posts based on interactions between online community users. The method of claim 1, The interaction includes any one or more of comments, comments, empathy, recommendations, trackbacks, scraps, links How to search online community posts based on interactions between online community users. A program for executing the method according to any one of claims 1 to 9 is recorded. Computer-readable recording media. An online community user information storage unit which stores first information about a plurality of online community users and a first score about the expertise of the online community users; A mutual information formed by an online community user different from the specific online community user among the online community users with respect to the second information about the online community post written by the online community user and the online community post written by a specific online community user among the online community users An online community post storage for storing third information about the action; A expertise calculation unit configured to calculate, for each of the on-line community users, a first score on the expertise of the specific online community user using the score on the expertise of the different online community users and the third information; A semantic similarity calculating unit that calculates a second score about semantic similarity between the input search word and the online community post, and A searcher for searching for the online community user or the online community post using the first score and the second score Interaction based online community post search device between online community users comprising a. The method of claim 11, The third information is the number of interactions, The specialty computing unit includes: a multiplier that performs a first operation for each of the different online community users, the multiplier performing multiplication of the third information and a score about the expertise of the different online community users, and the And an adder for performing a second operation including the summation of the results of the first operation. Online community post retrieval device based on interaction between online community users. The method of claim 11, Sorting the online community users in the order of the first scores and mapping a function having a higher value as the sorted ranking is closer to the middle rank, thereby obtaining a third score on the popularity of the particular online community user, respectively. It includes a popularity calculation unit for calculating, The search unit uses one or more of the first score or the third score and the second score. Online community post retrieval device based on interaction between online community users.

Images