• Journal of Korean Library and Information Science Society
• /
• v.46 no.1
• /
• pp.321-344
• /
• 2015

This paper presents a fusion approach to sentiment detection that combines multiple sources of evidence to retrieve blogs that contain opinions on a specific topic. Our approach to finding opinionated blogs on topic consists of first applying traditional information retrieval methods to retrieve blogs on a given topic and then boosting the ranks of opinionated blogs based on the opinion scores computed by multiple sentiment detection methods. Our sentiment detection strategy, whose central idea is to rely on a variety of complementary evidences rather than trying to optimize the utilization of a single source of evidence, includes High Frequency module, which identifies opinions based on the frequency of opinion terms (i.e., terms that occur frequently in opinionated documents), Low Frequency module, which makes use of uncommon/rare terms (e.g., "sooo good") that express strong sentiments, IU Module, which leverages n-grams with IU (I and you) anchor terms (e.g., I believe, You will love), Wilson's lexicon module, which uses a collection-independent opinion lexicon constructed from Wilson's subjectivity terms, and Opinion Acronym module, which utilizes a small set of opinion acronyms (e.g., imho). The results of our study show that combining multiple sources of opinion evidence is an effective method for improving opinion detection performance.

• Journal of Internet Computing and Services
• /
• v.21 no.6
• /
• pp.57-69
• /
• 2020

As more people share their opinions in online communities, such as Internet portals and social networking services, more opinions are manipulated for the benefit of particular individuals and groups. In particular, when manipulations occur for political purposes, they influence election results as well as government policies and the quality of life. This type of manipulation has targeted the general public, and their analysis and detection has also focused on such manipulation. However, to more efficiently spread propaganda, recent manipulations have targeted common interest groups(e.g., a group of those interested in real estate) and propagated information whose content and style are customized to those groups. This work characterizes such manipulations on common interest groups and proposes method to detect manipulations. To this end, we collected and analyzed opinions posted on 10 common interest groups before and after an election. As a result, we found that manipulations on common interest groups indeed occurred and were gradually increasing toward the election date. We also proposed a detection system that examines individual opinions, their authors, and their collaborators. Using the collected opinions, we demonstrated that the proposed system can accurately classify more than 90% of manipulated opinions and that many of these opinions were posted by multiple collaborators. We believe that regular audits of opinions using the proposed system can quickly isolate manipulations and decrease their impact. Moreover, the proposed features can be used to identify manipulations in domains other than politics.

• Journal of the Korea Society of Computer and Information
• /
• v.15 no.11
• /
• pp.21-29
• /
• 2010

Opinion retrieval which searches the opinions expressed in documents by users cannot outperform significantly yet traditional topical retrieval which searches the facts. Therefore, the focus of this paper is to identify the statistical characteristics which can be applied to opinion retrieval by comparing and analyzing the term statistics of opinion and non-opinion documents in the blog domain. The TREC Blogs06 collection and 150 TREC topics are used in the experiments. The difference between term probability distributions in opinion documents is measured by JS divergence, and the difference according to the topic types and topic domains is also investigated. Moreover, the term probabilities of opinion terms are analyzed comparatively. The main findings of this study include the following: it is necessary to consider the topic-specific characteristics for the opinion detection; it is effective to extract positive and negative opinion terms according to the topics; the topic types are complementary to the topic domains; and special attention has to be given to the usage of the positive opinion terms.

• Annual Conference on Human and Language Technology
• /
• 2012.10a
• /
• pp.58-61
• /
• 2012

본 논문에서는 트위터 사용자들의 의견을 바이어스 탐지 하기 위해, 핵심 자질 추출 방법으로 HITS 그래프를 이용한 방법을 제안한다. 제안하는 핵심 자질 추출 방법은 사람이 직접 추출하지 못하는 자질도 추출할 수 있는 장점을 보였다. 제안한 핵심 자질 추출이 바이어스 탐지에 유효함을 검증하기 위해 4개의 토픽에 대해 평가 했을 때 제안 모델이 기존 모델보다 우수한 성능을 보였다.

• The Journal of Society for e-Business Studies
• /
• v.18 no.4
• /
• pp.315-326
• /
• 2013

Users in social media post various types of opinions such as product reviews and movie reviews. It is a common trend that customers get assistance from the opinions in making their decisions. However, as opinion usage grows, distorted feedbacks also have increased. For example, exaggerated positive opinions are posted for promoting target products. So are negative opinions which are far from common evaluations. Finding these biased opinions becomes important to keep social media reliable. Techniques of opinion mining (or sentiment analysis) have been developed to determine sentiment polarity of opinionated documents. These techniques can be utilized for finding the biased opinions. However, the previous techniques have some drawback. They categorize the text into only positive and negative, and they also need a large amount of training data to build the classifier. In this paper, we propose methods for discovering the biased opinions which are skewed from the overall common opinions. The methods are based on angle based outlier detection and personalized PageRank, which can be applied without training data. We analyze the performance of the proposed techniques by presenting experimental results on a movie review dataset.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2008.05a
• /
• pp.1139-1142
• /
• 2008

인터넷의 이용이 증가함에 따라 네트워크를 통한 다양한 공격 역시 증가 추세에 있다. 따라서 네트워크 이상징후를 사전에 탐지하고 상황에 따라 유연하게 대처할 수 있도록 하기 위한 연구가 절실하다. 본 연구는 은닉마코프모델을 이용해 트래픽에서 이상징후를 탐지하는 기법을 제안한다. 제안하는 기법은 시계열 예측 기법을 이용해 트래픽에서 징후를 추출한다. 징후추출 과정의 결과를 은닉마코프모델을 활용한 징후판단과정을 통해 네트워크 이상징후인지를 판단하고 결정한다. 일련의 과정을 perl로 구현하고, 실제 공격이 포함된 트래픽을 사용하여 검증한다. 하지만 결과가 확연히 증명되지는 않는데, 이는 학습과정의 부족과 실제에 가까운 트래픽의 사용으로 인해 나타나는 현상으로 연구의 본질을 흐리지는 않는다고 판단된다. 오히려 실제 상황을 가정했을 때 접근이나 적용을 판단함에 관리자의 의견을 반영할 수 있으므로 공격의 탐지와 판단에 유연성을 증대시킬 수 있다. 본 연구는 실시간 네트워크의 상황 파악이나 네트워크에서의 신종 공격 탐지 및 분류에 응용가능할 것으로 기대된다.

• Journal of Internet Computing and Services
• /
• v.21 no.3
• /
• pp.29-39
• /
• 2020

Online communities, such as Internet portal sites and social media, have become popular since they allow users to share opinions and to obtain information anytime, anywhere. Accordingly, an increasing number of opinions are manipulated to the advantage of particular groups or individuals, and these opinions include falsified product reviews and political propaganda. Existing detection systems are built upon the characteristics of manipulated opinions for one particular time period. However, manipulation tactics change over time to evade detection systems and to more efficiently spread information, so detection systems should also evolve according to the changes. We therefore propose a system that helps observe and trace changes in manipulation tactics. This system classifies opinions into clusters that represent different tactics, and changes in these clusters reveal evolving tactics. We evaluated the system with over a million opinions collected during three election campaigns and found various changes in (i) the times when manipulations frequently occur, (ii) the methods to manipulate recommendation counts, and (iii) the use of multiple user IDs. We suggest that the operators of online communities perform regular audits with the proposed system to identify evolutions and to adjust detection systems.

• Proceedings of the Korean Information Science Society Conference
• /
• 2010.06c
• /
• pp.339-344
• /
• 2010

오늘날 인터넷 사용자들은 블로그나 뉴스 등의 매체에서 트랙백을 사용해 자신의 의견을 보다 자유롭게 나타낸다. 그러나 이러한 자유로움을 악용해 트랙백 스팸을 유발하여 네트워크의 자원을 낭비하고 방문자들에게 잘못된 정보를 전달해 해당 포스트의 신뢰를 떨어뜨린다. 트랙백 스팸은 유명한 포스트와 연계하여 자신의 포스트로 사용자들을 유도하는 특징을 가지기 때문에 일반적인 웹 스팸을 탐지하는 기술을 적용하기 어렵다. 따라서 본 논문에서는 자신이 작성한 글이 다른 사람의 글과 관련이 있다고 생각하여 다른사람의 글에 자신의 글을 링크시키는 트랙백의 특성을 이용하여 원본 페이지와 트랙백 페이지 그리고 트랙백 페이지의 아웃링크 내용상의 유사도와 동시 출현(co-occurrence) 정보를 이용하여 트랙백 스팸을 처리하고자 한다.

• The Korean Journal of Applied Statistics
• /
• v.26 no.6
• /
• pp.903-913
• /
• 2013

Social network services generate a considerable amount of social data every day on personal feelings or thoughts. This social data provides changing patterns of information production and consumption but are also a tool that reflects social phenomenon. We analyze negative emotional words from daily blogs to detect the change in blooger sentiment using multivariate control charts. We used the all the blogs produced between 1 January 2008 and 31 December 2009. Hotelling's T-square control chart control chart is commonly used to monitor multivariate quality characteristics; however, it assumes that quality characteristics follow multivariate normal distribution. The performance of a multivariate control chart is affected by this assumption; consequently, we introduce the support vector data description and its extension (K-control chart) suggested by Sun and Tsung (2003) and they are applied to detect the chage in blogger sentiment.

• Proceedings of the Korean Information Science Society Conference
• /
• 2007.10d
• /
• pp.112-117
• /
• 2007

사이버 침해란 정보시스템의 취약한 부분을 공격하여 시스템 내부에 침입하거나 시스템을 마비/파괴하는 등의 사고를 유발하는 모든 행위를 말한다. 이러한 사이버 침해의 피해를 줄이기 위해 국내외 많은 연구 기관과 업체에서는 침입탐지시스템과 같은 정보보호 기술을 연구 개발하여 상용화하고 있다. 그러나 기존의 정보보호 기술은 이미 발생한 침해를 탐지하여 피해의 확산을 막는 데만 한정적으로 사용되고, 침해의 발생 가능성을 예측하지는 못하기 때문에 점차 첨단화, 다양화되고 있는 사이버 침해에 대응하기 힘들다는 문제점을 갖는다. 본 논문에서는 보안 취약점을 이용한 사이버 침해를 대상으로 전문가 설문을 통해 사이버 침해의 발생 가능성을 예측하는 방법을 제안하고, 이를 위한 사이버 침해 예측 항목을 추출하였다. 예측 항목 추출은 3 단계로 구성되며, 첫 번째 단계에서는 기존 연구와 사례 분석을 통해 예측 항목의 계층 구조를 생성한다. 두 번째 단계에서는 첫 번째 단계를 통해 생성된 예측 항목들을 델파이 방법을 통해 개선하여 최적의 예측 항목을 결정한다. 마지막 단계에서는 각 항목들에 대한 쌍대 비교 설문을 진행하여 항목 간 가중치를 추출한다.