이미지 태그 간의 의미적 관계에 기반한 효율적인 태그 랭킹 기법

홍현기,정진우,이동호 한국정보과학회 2010 데이타베이스 연구 Vol.26 No.3

Folksonomy, considered as a core component for Web 2.0 user-participation architecture, is a classification system made by user's tags on the web resources. Recently, various approaches for image retrieval exploiting folksonomy have been proposed to improve the image search results. However, the characteristics of the tags such as semantic ambiguity, non-controlledness, and so on, limit the effectiveness of tags on image retrieval. Especially, tags associated with images in a random order do not provide any information about the relevance between a tag and an image. In this paper, we propose a novel image's tag ranking system which exploits the semantic relationships between tags for ordering the tags according to the relevance with an image. The proposed system consists of two phases: 1) tag propagation phase, 2) tag ranking phase. In tag propagation phase, for a given image without tags, after we perform preprocessing task for the tags of similar images, we collect the candidate tags with high occurrence frequency and propagate them to tag ranking phase. In tag ranking phase, for propagated tags, we estimate semantic relevance between tags using the semantic hierarchy relationship between concepts in WordNet. Afterward,we calculate ranking score of each tag by applying various weights to score of each tag. Finally, tags are ranked by ranking score of each tag, so that we can perform semantic relationship based tag ranking. From the experimental results on the Flickr photo collection over 10,000 images, we show the effectiveness of the proposed system. 웹 2.0의 사용자-참여형 아키텍쳐를 구성하는 핵심요소로 인식되고 있는 폭소노미는 웹 자원에 대하여 사용자의 태그에 의해 만들어지는 분류 시스템이다. 최근, 이미지 검색 결과를 개선하기 위하여 폭소노미를 활용하는 이미지 검색에 관한 다양한 시도들이 제안되고 있다. 하지만, 의미적 모호성, 규제를 받지 않는 점 등과 같은 태그의 특징들로 인해 이미지 검색시 태그의 실효성이 제한된다. 특히, 이미지와 관련된 태그들이 임의의 순서로 할당되어 있어 태그와 이미지 사이의 관련성에 대한 정보를 전혀 제공하지 못하고 있다. 본 논문에서는 태그들을 이미지와의 관련성에 따라 정렬하기 위해 태그간 의미적 관계를 활용하는 새로운 방식의 이미지 태그 랭킹 시스템을 제안한다. 제한된 시스템은 2개의 단계로 구성되어 있다: 1) 태그 전달 단계 2) 태그 랭킹 단계. 태그 전달 단계에서 태그가 없는 입력이미지가 주어지면, 입력이미지와 시각적으로 비슷한 이미지들의 태그들에 대해 텍스트 전처리 작업을 수행한 후, 높은 출현빈도를 보이는 후보 태그들을 수집하여 태그 랭킹 단계로 전달한다. 태그 랭킹 단계에서, 전달받은 태그들에 대해 WordNet에서 개념간 의미적 계층 관계를 이용하여 태그간 의미적 관련정도를 측정한다. 그 후, 각 태그의 점수에 다양한 가중치를 적용하여 각 태그의 랭킹점수를 계산한다. 태그들은 각 태그의 랭킹점수에 따라 정렬되어, 의미적 관계에 기반한 태그 랭킹을 수행할 수 있다. 10,000장 이상의 Flickr 이미지 모음에 대한 실험 결과를 통하여 제안된 시스템이 효과적임을 볼 수 있다.

Development of Tag Ranking-Based Image Search System

Si-Hwa Lee,Dae-Hoon Hwang,Sae-Hong Cho 보안공학연구지원센터 2014 International Journal of Software Engineering and Vol.8 No.2

다차원 인덱스를 위한 벡터형 태깅 연구

정재윤(Jae-Youn Jung),진현철(Hyeon-Cheol Zin),김종근(Chonggun Kim) 한국정보과학회 2009 정보과학회논문지 : 소프트웨어 및 응용 Vol.36 No.9

인터넷 사용이 대중화되면서 개인이 정보의 키워드 또는 검색할 주제에 따라 원하는 정보에 쉽게 접근할 수 있다. 이때 다양한 구조를 갖는 자료들의 속성을 잘 나타내는 메타데이터를 이용하면 검색의도에 보다 정확하게 부합하는 검색 결과를 얻을 수 있어 다양한 연구가 지속되고 있다. 본 연구는 소 그룹의 사용자들이 공동으로 관심 있는 웹 콘텐츠의 즐겨찾기를 공동으로 유지 관리하는 용도로 다차원 벡터형 태그를 제안한다. 제안하는 벡터형 태그는 정보 유용성을 나타내는 색인을 벡터방식으로 기술하고 이것을 활용해 정보의 분류·관리·재활용의 효율을 높이는 표현법이다. 벡터방식 태깅은 대상 키워드에 사용자들이 두 개 이상의 요소에 대한 우선순위를 부여하고 벡터 방식으로 표현한다. 이 때 벡터의 기본이 되는 벡터공간은 정보생성시간, 선호순위 등으로 구성한다. 벡터성분으로 산출할 수 있는 벡터크기가 정보의 유용성을 나타내며 순위측정의 기준이 된다. 제안방식에 의한 순위측정은 단순한 링크구조에 의해 측정된 순위와 비교하였을 때, 사용자의 검색의도에 부합하는 순위 정보를 제공하고 있다. A Internet user can easily access to the target information by web searching using some key-words or categories in the present Internet environment. When some meta-data which represent attributes of several data structures well are used, then more accurate result which is matched with the intention of users can be provided. This study proposes a multiple dimensional vector tagging method for the small web user group who interest in maintaining and sharing the bookmark for common interesting topics. The proposed method uses vector tag method for increasing the effect of categorization, management, and retrieval of target information. The vector tag composes with two or more components of the user defined priority. The basic vector space is created time of information and reference value. The calculated vector value shows the usability of information and became the metric of ranking. The ranking accuracy of the proposed method compares with that of a simply link structure, The proposed method shows better results for corresponding the intention of users.

A Folksonomy Ranking Framework: A Semantic Graph-based Approach

박현정,노상규 한국경영정보학회 2011 Asia Pacific Journal of Information Systems Vol.21 No.2

In collaborative tagging systems such as Delicious.com and Flickr.com, users assign keywords or tags to their uploaded resources, such as bookmarks and pictures, for their future use or sharing purposes. The collection of resources and tags generated by a user is called a personomy, and the collection of all personomies constitutes the folksonomy. The most significant need of the folksonomy users is to efficiently find useful resources or experts on specific topics. An excellent ranking algorithm would assign higher ranking to more useful resources or experts. What resources are considered useful in a folksonomic system? Does a standard superior to frequency or freshness exist? The resource recommended by more users with more expertise should be worthy of attention. This ranking paradigm can be implemented through a graph-based ranking algorithm. Two well-known representatives of such a paradigm are PageRank by Google and HITS(Hypertext Induced Topic Selection) by Kleinberg. Both PageRank and HITS assign a higher evaluation score to pages linked to more higher-scored pages. HITS differs from PageRank in that it utilizes two kinds of scores: authority and hub scores. The ranking objects of these pages are limited to Web pages, whereas the ranking objects of a folksonomic system are somewhat heterogeneous(i.e., users, resources, and tags). Therefore, uniform application of the voting notion of PageRank and HITS based on the links to a folksonomy would be unreasonable. In a folksonomic system, each link corresponding to a property can have an opposite direction, depending on whether the property is an active or a passive voice. The current research stems from the idea that a graph-based ranking algorithm could be applied to the folksonomic system using the concept of mutual interactions between entities, rather than the voting notion of PageRank or HITS. The concept of mutual interactions, proposed for ranking the Semantic Web resources, enables the calculation of importance scores of various resources unaffected by link directions. The weights of a property representing the mutual interaction between classes are assigned depending on the relative significance of the property to the resource importance of each class. This class-oriented approach is based on the fact that, in the Semantic Web, there are many heterogeneous classes; thus, applying a different appraisal standard for each class is more reasonable. This is similar to the evaluation method of humans, where different items are assigned specific weights, which are then summed up to determine the weighted average. We can check for missing properties more easily with this approach than with other predicate-oriented approaches. A user of a tagging system usually assigns more than one tags to the same resource, and there can be more than one tags with the same subjectivity and objectivity. In the case that many users assign similar tags to the same resource, grading the users differently depending on the assignment order becomes necessary. This idea comes from the studies in psychology wherein expertise involves the ability to select the most relevant information for achieving a goal. An expert should be someone who not only has a large collection of documents annotated with a particular tag, but also tends to add documents of high quality to his/her collections. Such documents are identified by the number, as well as the expertise, of users who have the same documents in their collections. In other words, there is a relationship of mutual reinforcement between the expertise of a user and the quality of a document. In addition, there is a need to rank entities related more closely to a certain entity. Considering the property of social media that ensures the popularity of a topic is temporary, recent data should have more weight than old data. We propose a comprehensive folksonomy ranking framework in which all these considerations are dealt with and that can be easily... In collaborative tagging systems such as Delicious.com and Flickr.com, users assign keywords or tags to their uploaded resources, such as bookmarks and pictures, for their future use or sharing purposes. The collection of resources and tags generated by a user is called a personomy, and the collection of all personomies constitutes the folksonomy. The most significant need of the folksonomy users is to efficiently find useful resources or experts on specific topics. An excellent ranking algorithm would assign higher ranking to more useful resources or experts. What resources are considered useful in a folksonomic system? Does a standard superior to frequency or freshness exist? The resource recommended by more users with more expertise should be worthy of attention. This ranking paradigm can be implemented through a graph-based ranking algorithm. Two well-known representatives of such a paradigm are PageRank by Google and HITS(Hypertext Induced Topic Selection) by Kleinberg. Both PageRank and HITS assign a higher evaluation score to pages linked to more higher-scored pages. HITS differs from PageRank in that it utilizes two kinds of scores: authority and hub scores. The ranking objects of these pages are limited to Web pages, whereas the ranking objects of a folksonomic system are somewhat heterogeneous(i.e., users, resources, and tags). Therefore, uniform application of the voting notion of PageRank and HITS based on the links to a folksonomy would be unreasonable. In a folksonomic system, each link corresponding to a property can have an opposite direction, depending on whether the property is an active or a passive voice. The current research stems from the idea that a graph-based ranking algorithm could be applied to the folksonomic system using the concept of mutual interactions between entities, rather than the voting notion of PageRank or HITS. The concept of mutual interactions, proposed for ranking the Semantic Web resources, enables the calculation of importance scores of various resources unaffected by link directions. The weights of a property representing the mutual interaction between classes are assigned depending on the relative significance of the property to the resource importance of each class. This class-oriented approach is based on the fact that, in the Semantic Web, there are many heterogeneous classes; thus, applying a different appraisal standard for each class is more reasonable. This is similar to the evaluation method of humans, where different items are assigned specific weights, which are then summed up to determine the weighted average. We can check for missing properties more easily with this approach than with other predicate-oriented approaches. A user of a tagging system usually assigns more than one tags to the same resource, and there can be more than one tags with the same subjectivity and objectivity. In the case that many users assign similar tags to the same resource, grading the users differently depending on the assignment order becomes necessary. This idea comes from the studies in psychology wherein expertise involves the ability to select the most relevant information for achieving a goal. An expert should be someone who not only has a large collection of documents annotated with a particular tag, but also tends to add documents of high quality to his/her collections. Such documents are identified by the number, as well as the expertise, of users who have the same documents in their collections. In other words, there is a relationship of mutual reinforcement between the expertise of a user and the quality of a document. In addition, there is a need to rank entities related more closely to a certain entity. Considering the property of social media that ensures the popularity of a topic is temporary, recent data should have more weight than old data. We propose a comprehensive folksonomy ranking framework in which all these considerations are dealt with and that can be easily cu...

소셜 북마킹 시스템에서의 북마크와 태그 정보를 활용한 웹 콘텐츠 랭킹 알고리즘

박수진(Su-Jin Park),이시화(Si-Hwa Lee),황대훈(Dae-Hoon Hwang) 한국멀티미디어학회 2010 멀티미디어학회논문지 Vol.13 No.8

현재 웹 2.0 환경에서의 핵심 기술 중 하나는 사용자가 관심 있는 웹페이지를 태깅 및 북마킹 하는 소설 북마킹 기술이다. 소셜 북마킹은 웹 콘텐츠에 태깅된 북마크 정보 및 태깅 결과를 기반으로 검색, 분류, 공유를 통해 효율적인 정보 제공을 주목적으로 하고 있다. 그러나 현재 소셜 북마킹 시스템들은 웹 콘텐츠의 사용자들의 관심 정도를 측정할 수 있는 북마크 수 및 검색과 분류를 목적으로 하는 태그 정보를 각각 독립적으로 검색에 활용하는 방식을 사용하고 있다. 이는 소설 북마킹 시스템에서 중요한 특징을 가지는 북마크와 태깅 기술을 효율적으로 활용하지 못하는 결과가 된다. 이에 본 연구에서는 태그 클러스터링을 통한 연관 태그 추출에 관한 선행연구를 기반으로, 북마크 정보와 혼합하기 위한 웹 콘텐츠 랭킹 알고리즘을 제안하였다. 또한 제안 알고리즘의 효율성 분석을 위해 기존 검색 방법론들과의 비교평가를 시행하였으며, 그 결과 본 연구의 핵심적인 특징인 북마크와 태그 정보를 함께 활용한 소셜 북마크 시스템이 기존 시스템보다 효율적인 검색결과를 도출하였다. In current Web 2.0 environment, one of the most core technology is social bookmarking which users put tags and bookmarks to their interesting Web pages. The main purpose of social bookmarking is an effective information service by use of retrieval, grouping and share based on user's bookmark information and tagging result of their interesting Web pages. But, current social bookmarking system uses the number of bookmarks and tag information separately in information retrieval, where the number of bookmarks stand for user's degree of interest on Web contents, information retrieval. and classification serve the purpose of tag information. Because of above reason, social bookmarking system does not utilize effectively the bookmark information and tagging result. This paper proposes a Web contents ranking algorithm combining bookmarks and tag information, based on preceding research on associative tag extraction by tag clustering. Moreover, we conduct a performance evaluation comparing with existing retrieval methodology for efficiency analysis of our proposed algorithm. As the result, social bookmarking system utilizing bookmark with tag, key point of our research, deduces a effective retrieval results compare with existing systems.

태그쌍의 의미유사도 기반 태그 랭킹 시스템

이시화(Si-Hwa Lee),황대훈(Dae-Hoon Hwang) 한국멀티미디어학회 2013 멀티미디어학회논문지 Vol.16 No.11

순위 조작 방지를 위한 소설 태깅 서비스 설계

정한영(Han-Young Jung),최옥경(Okkyung Choi),예홍진(Hongjin Yeh) 한국컴퓨터정보학회 2012 한국컴퓨터정보학회 학술발표논문집 Vol.20 No.2

최근 소설 네트워크 서비스(Social Netwrok Services)를 활용한 소설 태깅 서비스에 대한 연구가 활발하게 진행 중이다. 특히 태그 기술을 이용한 협력적 태깅 시스템(collaborative tagging system)은 북마크, 문서, 사진, 동영상과 같은 웹 자원을 조직화하고 공유할 수 있는 수단으로 제공하고 있다. 그러나 광고 홍보 목적을 가진 스패머들은 콘텐츠와 관련 없는 태그를 달아 놓아 검색 키워드와 무관한 결과 값이 검색되어 웹 검색 서비스를 이용하는 사용자들에게 불편을 주고 있다. 따라서 본 연구에서는 콘텐츠와 태그의 연관성을 높이기 위해 태그에 일반 사용자가 유사도를 입력할 수 있는 추천 시스템을 적용하여 순위 조작 방지방법을 제안하였다.

블로그의 구조적 특성을 고려한 효율적인 블로그 검색 알고리즘

김정훈(Jung-Hoon Kim),윤태복(Tae-Bok Yoon),이지형(Jee-Hyong Lee) 한국정보과학회 2009 정보과학회논문지 : 소프트웨어 및 응용 Vol.36 No.7

오늘날, 대부분의 웹 페이지는 블로그영역에서 생성되고 기존의 웹 페이지 또한 블로그영역으로 전환되어가고 있다. 블로그 페이지는 트랙백연결, 블로거, 태그, 댓글과 같은 기존 웹 페이지에는 존재하지 않는 특징이 있다. 따라서 이러한 차이를 반영하지 않는 전통적인 웹 페이지 랭킹 알고리즘을 블로그 페이지에 단순히 적용하는 것은 효율적인 검색을 위해 적절하지 않다. 본 논문에서는 이러한 문제를 해결하기 위해 블로그 검색을 위한 “블로그-랭크” 알고리즘을 제안한다. 제안하는 알고리즘은 블로그의 구조적특징들을 활용하여 트랙백 연결성, 블로거의 명성, 사용자 반응성을 평가하고 이를 기반으로 블로그 페이지를 랭크 한다. 우리는 알고리즘의 검색효율성을 증명하기 위해 제안한 알고리즘을 적용한 블로그 검색 시스템을 구현하고 기존의 블로그 검색시스템과 검색효율성을 비교하였으며, 그 결과 블로그-랭크 알고리즘을 적용한 검색시스템이 기존의 검색시스템보다 더욱 뛰어난 검색효율성을 보임을 확인하였다. Today, most web pages are being created in the blogspace or evolving into the blogspace. A blog entry (blog page) includes non-traditional features of Web pages, such as trackback links, bloggers' authority, tags, and comments. Thus, the traditional rank algorithms are not proper to evaluate blog entries because those algorithms do not consider the blog specific features. In this paper, a new algorithm called "Blog-Rank" is proposed. This algorithm ranks blog entries by calculating bloggers' reputation scores, trackback scores, and comment scores based on the features of the blog entries. This algorithm is also applied to searching for information related to the users’ queries in the blogspace. The experiment shows that it finds the much more relevant information than the traditional ranking algorithms.

Timeline Tag Cloud Generation for Broadcasting Contents using Blog Postings

손정우(Jeong-Woo Son),김화숙(Hwa-Suk Kim),김선중(Sun-Joong Kim),조기성(Keeseong Cho) Korean Institute of Information Scientists and Eng 2015 정보과학회논문지 Vol.42 No.5

A Folksonomy Ranking Framework : A Semantic Graph-based Approach

Hyunjung Park,Sangkyu Rho 한국경영정보학회 2011 Asia Pacific Journal of Information Systems Vol.21 No.2