바로가기메뉴

본문 바로가기 주메뉴 바로가기

logo

검색어: fusion, 검색결과: 4
초록보기
초록

Abstract

This paper describes a Web search optimization study that investigates both static and dynamic tuning methods for optimizing system performance. We extended the conventional fusion approach by introducing the “dynamic tuning” process with which to optimize the fusion formula that combines the contributions of diverse sources of evidence on the Web. By engaging in iterative dynamic tuning process, where we successively fine-tuned the fusion parameters based on the cognitive analysis of immediate system feedback, we were able to significantly increase the retrieval performance.Our results show that exploiting the richness of Web search environment by combining multiple sources of evidence is an effective strategy.

2
백우진(건국대학교) ; Sun-Eun Jung(Konkuk U) ; Euigun Ahn(Yonsei U) ; 김기용(건국대학교) ; 신문선(건국대학교) 2008, Vol.25, No.2, pp.49-68 https://doi.org/10.3743/KOSIM.2008.25.2.049
초록보기
초록

Abstract

In many information retrieval experiments, the data fusion techniques have been used to achieve higher effectiveness in comparison to the single evidence-based retrieval. However, there had not been many image retrieval studies using the data fusion techniques especially in combining retrieval results based on multiple retrieval methods. In this paper, we describe how the image retrieval effectiveness can be improved by combining two sets of the retrieval results using the Sobel operator-based edge detection and the Self Organizing Map(SOM) algorithms. We used the clip art images from a commercial collection to develop a test data set. The main advantage of using this type of the data set was the clear cut relevance judgment, which did not require any human interven- tion.

3
Yang, Kiduk(경북대학교) ; Lokman Meho(American University of Beirut, Lebanon) 2011, Vol.28, No.2, pp.79-96 https://doi.org/10.3743/KOSIM.2011.28.2.079
초록보기
초록

Abstract

Despite the widespread use, critics claim that citation analysis has serious limitations in evaluating the research performance of scholars. First, conventional citation analysis methods yield one-dimensional and sometimes misleading evaluation as a result of not taking into account differences in citation quality, not filtering out citation noise such as self-citations, and not considering non-numeric aspects of citations such as language, culture, and time. Second, the citation database coverage of today is disjoint and incomplete, which can result in conflicting quality assessment outcomes across different data sources. This paper discuss the findings from a citation analysis study that measured the impact of scholarly publications based on the data mined from Web of Science, Scopus, and Google Scholar, and briefly describes a work-in-progress prototype system called CiteSearch, which is designed to overcome the weaknesses of existing citation analysis methods with a robust citation-based quality assessment approach.

4
이용구(계명대학교) ; 정영미(연세대학교) 2007, Vol.24, No.1, pp.321-342 https://doi.org/10.3743/KOSIM.2007.24.1.321
초록보기
초록

이 연구에서는 수작업 태깅없이 기계가독형 사전을 이용하여 자동으로 의미를 태깅한 후 학습데이터로 구축한 분류기에 대해 의미를 분류하는 단어 중의성 해소 모형을 제시하였다. 자동 태깅을 위해 사전 추출 정보 기반 방법과 연어 공기 기반 방법을 적용하였다. 실험 결과, 자동 태깅에서는 복수 자질 축소를 적용한 사전 추출 정보 기반 방법이 70.06%의 태깅 정확도를 보여 연어 공기 기반 방법의 56.33% 보다 24.37% 향상된 성능을 가져왔다. 사전 추출 정보 기반 방법을 이용한 분류기의 분류 정학도는 68.11%로서 연어 공기 기반 방법의 62.09% 보다 9.7% 향상된 성능을 보였다. 또한 두 자동 태깅 방법을 결합한 결과 태깅 정확도는 76.09%, 분류 정확도는 76.16%로 나타났다.

Abstract

This study presents an effective word sense disambiguation model that does not require manual sense tagging process by automatically tagging the right sense using a machine-readable dictionary, and attempts to classify the senses of those words using a classifier built from the training data. The automatic tagging technique was implemnted by the dictionary information-based and the collocation co-occurrence-based methods. The dictionary information-based method that applied multiple feature selection showed the tagging accuracy of 70.06%, and the collocation co-occurrence-based method 56.33%. The sense classifier using the dictionary information-based tagging method showed the classification accuracy of 68.11%, and that using the collocation co-occurrence-based tagging method 62.09%. The combined tagging method applying data fusion technique achieved a greater performance of 76.09% resulting in the classification accuracy of 76.16%.

정보관리학회지