• Proceedings of the Korean Information Science Society Conference
• /
• 2010.06c
• /
• pp.291-294
• /
• 2010

의견의 주체를 찾는 일은 의견 분석의 결과를 활용 하는데 있어 필수적인 분야이다. 본 논문은 발안자를 찾는 시스템의 성능을 높이기 위해 이전논문에 제안하였던 단어에 의견주체의 후보로서의 점수를 부여하는 방법을 개선하였고 미등록어 문제를 해결하기 위해 taxonomy에 의존하여 기존단어의 점수를 이용하는 방법을 제안하였다. 본 논문에서 제안한 방법은 Baseline과 비교하여 F1값이 18.9% 증가하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2009.11a
• /
• pp.337-338
• /
• 2009

블로그문서의 의견극성분류 연구는 주로 기계학습기법에 기반한 방법이었고, 이때 주로 활용된 자질은 명사, 동사 등의 품사정보와 의견어 어휘정보였다. 하지만 하나의 의견어 어휘만을 고려한다면 그 극성을 판별하는데 필요한 정보가 충분하지 않아 부정확한 결과를 도출하는 경우가 발생할 수 있다. 본 논문에서는 여러 어휘를 동시에 고려하였을 때 보다 정확한 의견분류를 수행할 수 있을 것이라는 가정을 세웠다. 본 논문에서는 효과적인 의견어휘자질의 추출을 위하여 의견이 내포될 가능성이 높은 의견어휘를 기반으로 의존구문분석을 통해 의존트리패턴을 추출하였고, 제안하는 PF-IDF가중치를 적용하여 지지벡터기계(SVM)와 다항시행접근 단순베이지안(MNNB)알고리즘으로 비교 실험을 수행하였다. 기준시스템인 TF-IDF가중치 기법에 비해 정확도(accuracy)가 지지벡터기계에서 5%, 다항시행접근 단순베이지안에서 8.9% 향상된 성능을 보였다.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.21 no.1
• /
• pp.68-79
• /
• 2011

This paper focuses on the opinion information extraction and analysis system through Web mining that is based on statistics collected from Web contents. That is, users' opinion information which is scattered across several websites can be automatically analyzed and extracted. The system provides the opinion information search service that enables users to search for real-time positive and negative opinions and check their statistics. Also, users can do real-time search and monitoring about other opinion information by putting keywords in the system. Proposing technique proved that the actual performance is excellent by comparison experiment with other techniques. Performance evaluation of function extracting positive/negative opinion information, the performance evaluation applying dynamic window technique and tokenizer technique for multilingual information retrieval, and the performance evaluation of technique extracting exact multilingual phonetic translation are carried out. The experiment with typical movie review sentence and Wikipedia experiment data as object as that applying example is carried out and the result is analyzed.

• Journal of the Korean Society for information Management
• /
• v.25 no.4
• /
• pp.115-129
• /
• 2008

Recently, as its growth and popularization, the Web is changed into the place where people express, share and debate their opinions rather than the space of information seeking. Accordingly, the needs for searching opinions expressed in the Web are also increasing. However, it is difficult to meet these needs by using a classical information retrieval system that only concerns the relevance between the user's query and documents. Instead, a more advanced system that captures subjective information through documents is required. The proposed system effectively retrieves opinionated documents by utilizing an existing information retrieval system. This paper proposes a kind of hybrid method which can utilize both a dictionary-based opinion analysis technique and a machine learning based opinion analysis technique. Experimental results show that the proposed method is effective in improving the performance.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2015.04a
• /
• pp.1001-1004
• /
• 2015

소설미디어의 급증과 함께 온라인 리뷰의 의존성이 급증하는 가운데 사용자의 올바른 의사결정을 저해하는 기만적 의견 스팸 이슈가 새롭게 주목받고 있다. 기존의 의견 스팸 연구는 실제 리뷰와 의견 스팸 간의 차이를 어휘, 품사 또는 감정단어와 같은 표면적 자질을 통해 설명하였으나 그들간의 의미적 연결관계는 고려하지 않았다. 본 논문에서는 1) 의미적 프레임 기반의 텍스트 분석기법을 제안하고, 이를 바탕으로 2) 의견 스팸과 실제 리뷰간의 의미적 차이가 있음을 규명하며 3) 새로운 의미적 프레임 자질을 사용하여 기존의 의견 스팸 분류 성능을 향상시킬 수 있음을 보인다.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.11 no.8
• /
• pp.3070-3075
• /
• 2010

Through Web 2.0 days, the end users express their opinions and thoughts for blogs and community spaces on the Internet. These opinions and thoughts are used to purchase products, however, users only refer to a few comments not overall opinions. Opinion Analysis System is an opinion search, developed from a natural language search, which analyzes the product's positive or negative evaluations using opinions of products and services on the Internet. In this paper, we suggest a syntactic analysis and inverse processing system that studies and processes 'Positive', 'Negative', 'Neutral' in addition to 'Inverse' information to analyze 'positive' or 'negative' for the core of sentences in Opinion Analysis Service.

• Journal of KIISE:Databases
• /
• v.36 no.4
• /
• pp.254-262
• /
• 2009

As the trend of e-business activities develop, customers come into contact with products through on-line shopping sites and lots of customers refer product reviews before the purchasing on-line. However, as the volume of product reviews grow, it takes a great deal of time and effort for customers to read and evaluate voluminous product reviews. Lately, attention is being paid to Opinion Mining(OM) as one of the effective solutions to this problem. In this paper, we propose an efficient method for opinion sentiment classification of product reviews using product specific context information of words occurred in the reviews. We define the context information of words and propose the application of context for sentiment classification and we show the performance of our method through the experiments. Additionally, in case of word corpus construction, we propose the method to construct word corpus automatically using the review texts and review scores in order to prevent traditional manual process. In consequence, we can easily get exact sentiment polarities of opinion words in product reviews.

• 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
• /
• 2009.11a
• /
• pp.725-726
• /
• 2009

오늘날 인터넷의 발달 때문에 인터넷으로 쉽게 신문을 볼 수 있게 되었다. 또한, 해당 기사에 대한 의견을 사용자끼리 쉽게 교환할 수 있다. 본 논문에서는 이러한 인터넷 기사의 사용자 의견들에 Opinion Mining 기술을 활용하여 해당 기사 대상의 특징을 올바르게 파악하는 방법을 제안한다.

• KIISE Transactions on Computing Practices
• /
• v.20 no.9
• /
• pp.492-497
• /
• 2014

In this paper, we propose a method of retrieving opinioned tweets in Twitter, which is the one of the popular Social Network Services and shares diverse opinions among various users. In typical opinion retrieval systems, they may consider the presence of sentiment phrases (subjectivity) as the important factor even if the subjective phrases are not related to a given query or speaker. To alleviate these problems, we utilized the syntactic structure of a sentence to identify the relationships between 1) subjectivity-query and 2) subjectivity-speaker and 3) the syntactic role of subjectivity. Besides, our learning-to-rank approach is trained to retrieve opinioned tweets based on query-relevance, textual features, user information, and Twitter-specific features. Experimental results on real world data show that our proposed method can achieve better performance than several baseline methods in terms of precision and nDCG.