• 전자공학회논문지CI
• /
• 제46권5호
• /
• pp.14-28
• /
• 2009

PageRank 알고리즘은 구글(Google)등의 검색 엔진에서 웹 페이지의 순위(rank)를 정하는 중요한 요소이다. PageRank 알고리즘의 순위 품질(ranking quality)을 향상시키기 위해 많은 변형 알고리즘들이 제안되었지만 어떤 변형 알고리즘(혹은 변형 알고리즘들간의 조합)이 가장 좋은 순위 품질을 제공하는지가 명확하지 않다. 본 논문에서는 PageRank 알고리즘의 잘 알려진 변형 알고리즘들과 그들 간의 조합들에 대해 순위 품질을 평가한다. 이를 위해, 먼저 변형 알고리즘들을 웹의 링크(link) 구조를 이용하는 링크기반 방법(Link-based approaches)과 웹의 의미 정보를 이용하는 지식기반 방법(Knowledge-based approaches)으로 분류한다. 다음으로, 이 두 가지 방법에 속하는 알고리즘들을 조합한 알고리즘들을 제안하고, 변형 알고리즘들과 그들을 조합한 알고리즘들을 구현한다. 백만 개의 웹 페이지들로 구성된 실제 데이터에 대한 실험을 통해 PageRank의 변형 알고리즘들과 그들 간의 조합들로부터 가장 좋은 순위 품질을 제공하는 알고리즘을 찾는다.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2003년도 가을 학술발표논문집 Vol.30 No.2 (1)
• /
• pp.169-171
• /
• 2003

웹에서의 검색 방법에는 크게 Text-Based 기법과 Link-Based 기법이 있다. 본 논문은 그 중에서 Link-Based 기법의 하나인 PageRank 알고리즘에 대해 연구 하고자 한다. 이 PageRank 알고리즘은 각 페이지의 중요성을 수치로 계산하는 방법이다. 하지만 이 알고리즘에서는 페이지에서 페이지로 링크를 따라갈 확률의 값을 일정하게 주어서 모든 페이지의 값을 획일적으로 계산하였기 때문에 각 페이지의 검색 효율성에 문제가 있다고 판단하여, 이를 해결하고자 본 논문은 페이지사이의 유사도를 측정하여 유사도에 따라 링크를 따라가는 확률 값인 Damping factor값을 다르게 부여하여 검색의 효율성을 높였다. 이를 위하여 두 가지 방법의 실험을 통하여 구현, 증명하였다.

• 정보과학회 논문지
• /
• 제44권2호
• /
• pp.179-185
• /
• 2017

문서의 요약은 요약문 내의 문장들끼리 서로 연관성 있게 이어져야 하고 하나의 짜임새 있는 글이 되어야 한다. 본 논문에서는 위의 목적을 달성하기 위해 문장 간의 유사도와 수반 관계(Entailment)를 고려하여 문서 내에서 연관성이 크고 의미, 개념적인 연결성이 높은 문장들을 추출할 수 있도록 하였다. 본 논문에서는 Recurrent Neural Network 기반의 문장 관계 추론 모델과 그래프 기반의 랭킹(Graph-based ranking) 알고리즘을 혼합하여 단일 문서 추출요약 작업에 적용한 새로운 알고리즘인 TextRank-NLI를 제안한다. 새로운 알고리즘의 성능을 평가하기 위해 기존의 문서요약 알고리즘인 TextRank와 동일한 데이터 셋을 사용하여 성능을 비교 분석하였으며 기존의 알고리즘보다 약 2.3% 더 나은 성능을 보이는 것을 확인하였다.

• 한국정보시스템학회지:정보시스템연구
• /
• 제33권3호
• /
• pp.105-121
• /
• 2024

Purpose The purpose of this paper is to develop a method to improve topic representation by incorporating the TextRank technique in Bertopic-based topic modeling and additional indicators for determining the optimal number of topics. Design/methodology/approach In this paper, we propose a method to extract important documents from documents assigned to each topic of a topic model using the TextRank technique, and to calculate secondary diversity and generate topic representations based on the results. First, we integrate the TextRank algorithm into the Bertopic-based topic modeling process to set local secondary labels for each topic. The secondary labels of each topic are derived through extractive summarization based on the TextRank algorithm. Second, we improve the accuracy of selecting the optimal number of topics by calculating the secondary diversity index based on the extractive summary results of each topic. Third, we improve the efficiency by utilizing ChatGPT when deriving the labels of each topic. Findings As a result of performing case analysis and analysis evaluation using the proposed method, it was confirmed that topic representation based on TextRank results generated more accurate topic labels and that the secondary diversity index was a more effective index for determining the optimal number of topics.

• 디지털콘텐츠학회 논문지
• /
• 제19권1호
• /
• pp.157-163
• /
• 2018

SNS를 이용하는 사람들은 매일 자신의 다양한 생각을 SNS에 게시한다. SNS에 게시된 데이터는 수많은 사람들의 생각과 의견이 담겨있다고 할 수 있다. 특히 트위터에서 서비스되는 인기 화제어는 사용자가 올린 글에서 자주 등장한 단어의 횟수를 집계해 순위를 결정한다. 하지만 이와 같은 방법은 단순히 중복된 단어가 나열된 불필요한 데이터에 민감하다. 제안하는 방법은 단어간의 관계도를 이용한 단어의 화제성을 기반으로 순위를 결정하므로 불필요한 데이터의 영향을 적게 받고 주요단어를 안정적으로 추출할 수 있다. 성능 비교를 위하여 내림차순 화제어 순위와 상위 20개중에서 의미 없는 화제어의 비율 측면에서 형태소 분석과 PageRank 기반의 제안 방식과 단순 등장 횟수 기반의 기존 방식을 비교한다. 제안하는 방안과 기존 방안은 상위 20개중에서 무의미한 화제어를 각각 55%과 70%를 순위권에 포함시켰으며 제안한 방법이 기존 방법과 비교할 때 15% 정도 향상된다.

• 복식
• /
• 제58권5호
• /
• pp.102-117
• /
• 2008

This study shows the badge system for military officials of Joseon dynasty. The badge system for military officials of the 15th century consists of rank badges with tiger and leopard for the first and second ranks and rank badges with bear for the third rank. According to the code of laws, military officials are supposed to wear the rank badges with four different kinds of animals in Joseon dynasty. However, the badge system shown in the code of laws sometimes does not match with the badges in practices. Based on the literature, remaining badges and the badges in portraits, six different kinds of badges with animals are found : First, rank badges with tiger and leopard were used until the late 16th century. Second, rank badges with tiger were found in the period between the early 17th century and the latter 18th century. Third, rank badges with Haechi were found in the early 17th century. Fourth, rank badges with lions can be found in remains of the mid 17th century, the literature and the portrait of the late 18th century. Finally, the rank badges with double leopards or with single leopard were found from a portrait dated the late of 18th century to the last period of Joseon dynasty.

• Communications for Statistical Applications and Methods
• /
• 제10권3호
• /
• pp.1007-1015
• /
• 2003

In this paper we introduced a new score generating function for the rank dispersion function in a general linear model. Based on the new score function, we derived the null asymptotic theory of the rank-based hypothesis testing in a linear model. In essence we showed that several rank test statistics, which are primarily focused on our new score generating function and new dispersion function, are mainly distribution free and asymptotically converges to a chi-square distribution.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권1호
• /
• pp.109-134
• /
• 2018

Collaborative Filtering (CF) is widely used in recommendation field, which can be divided into rating-based CF and learning-to-rank based CF. Although many methods have been proposed based on these two kinds of CF, there still be room for improvement. Firstly, the data sparsity problem still remains a big challenge for CF algorithms. Secondly, the malicious rating given by some illegal users may affect the recommendation accuracy. Existing CF algorithms seldom took both of the two observations into consideration. In this paper, we propose a recommendation method based on listwise learning-to-rank by incorporating users' social information. By taking both ratings and order of items into consideration, the Plackett-Luce model is presented to find more accurate similar users. In order to alleviate the data sparsity problem, the improved matrix factorization model by integrating the influence of similar users is proposed to predict the rating. On the basis of exploring the trust relationship between users according to their social information, a listwise learning-to-rank algorithm is proposed to learn an optimal ranking model, which can output the recommendation list more consistent with the user preference. Comprehensive experiments conducted on two public real-world datasets show that our approach not only achieves high recommendation accuracy in relatively short runtime, but also is able to reduce the impact of malicious ratings.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제10권2호
• /
• pp.146-151
• /
• 2010

This paper proposes a rank-based control method of mutation probability for improving the performances of genetic algorithms (GAs). In order to improve the performances of GAs, GAs should not fall into premature convergence phenomena and should also be able to easily get out of the phenomena when GAs fall into the phenomena without destroying good individuals. For this, it is important to keep diversity of individuals and to keep good individuals. If a method for keeping diversity, however, is not elaborately devised, then good individuals are also destroyed. We should devise a method that keeps diversity of individuals and also keeps good individuals at the same time. To achieve these two objectives, we introduce a rank-based control method of mutation probability in this paper. We set high mutation probabilities to lowly ranked individuals not to fall into premature convergence phenomena by keeping diversity and low mutation probabilities to highly ranked individuals not to destroy good individuals. We experimented our method with typical four function optimization problems in order to measure the performances of our method. It was found from extensive experiments that the proposed rank-based control method could accelerate the GAs considerably.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권1호
• /
• pp.1-19
• /
• 2021

The graph-based tag recommendation algorithm FolkRank can effectively utilize the relationships between three entities, namely users, items and tags, and achieve better tag recommendation performance. However, FolkRank does not consider the internal relationships of user-user, item-item and tag-tag. This leads to the failure of FolkRank to effectively map the tagging behavior which contains user neighbors and item neighbors to a tripartite graph. For item-item relationships, we can dig out items that are very similar to the target item, even though the target item may not have a strong connection to these similar items in the user-item-tag graph of FolkRank. Hence this paper proposes an improved FolkRank algorithm named FolkRank++, which fully considers the user-user and item-item internal relationships in tag recommendation by adding the correlation information between users or items. Based on the traditional FolkRank algorithm, an initial weight is also given to target user and target item's neighbors to supply the user-user and item-item relationships. The above work is mainly completed from two aspects: (1) Finding items similar to target item according to the attribute information, and obtaining similar users of the target user according to the history behavior of the user tagging items. (2) Calculating the weighted degree of items and users to evaluate their importance, then assigning initial weights to similar items and users. Experimental results show that this method has better recommendation performance.