초록

개체들 사이의 관계를 저차원 공간에 매핑하는 다차원척도법을 수행하기 위한 다양한 방법과 알고리즘이 개발되어왔다. 그러나 PROXSCAL이나 ALSCAL과 같은 기존의 기법들은 50개 이상의 개체를 포함하는 데이터 집합을 대상으로 개체 간의 관계와 군집 구조를 시각화하는데 있어서 효과적이지 못한 것으로 나타났다. 이 연구에서 제안하는 군집 지향 척도법 CLUSCAL(CLUster-oriented SCALing)은 기존 방법과 달리 입력되는 데이터의 군집 구조를 고려하도록 고안되었다. 50명의 저자동시인용 데이터와 85개 단어의 동시출현 데이터에 대해서 적용해본 결과 제안한 CLUSCAL 기법은 군집 구조를 잘 식별할 수 있는 MDS 지도를 생성하는 유용한 기법임이 확인되었다.

Abstract

There have been many methods and algorithms proposed for multidimensional scaling to mapping the relationships between data objects into low dimensional space. But traditional techniques, such as PROXSCAL or ALSCAL, were found not effective for visualizing the proximities between objects and the structure of clusters of large data sets have more than 50 objects. The CLUSCAL(CLUster-oriented SCALing) technique introduced in this paper differs from them especially in that it uses cluster structure of input data set. The CLUSCAL procedure was tested and evaluated on two data sets, one is 50 authors co-citation data and the other is 85 words co-occurrence data. The results can be regarded as promising the usefulness of CLUSCAL method especially in identifying clusters on MDS maps.

초록

Abstract

As co-authorship has been prevalent within science communities, counting the credit of co-authors appropriately is an important consideration, particularly in the context of identifying the knowledge structure of fields with author-based analysis. The purpose of this study is to compare the characteristics of co-author credit counting methods by utilizing correlations, multidimensional scaling, and pathfinder networks. To achieve this purpose, this study analyzed a dataset of 2,014 journal articles and 3,892 cited authors from the Journal of the Architectural Institute of Korea: Planning & Design from 2003 to 2008 in the field of Architecture in Korea. In this study, six different methods of crediting co-authors are selected for comparative analyses. These methods are first-author counting (m1), straight full counting (m2), and fractional counting (m3), proportional counting with a total score of 1 (m4), proportional counting with a total score between 1 and 2 (m5), and first-author-weighted fractional counting (m6). As shown in the data analysis, m1 and m2 are found as extreme opposites, since m1 counts only first authors and m2 assigns all co-authors equally with a credit score of 1. With correlation and multidimensional scaling analyses, among five counting methods (from m2 to m6), a group of counting methods including m3, m4, and m5 are found to be relatively similar. When the knowledge structure is visualized with pathfinder network, the knowledge structure networks from different counting methods are differently presented due to the connections of individual links. In addition, the internal validity shows that first-author-weighted fractional counting (m6) might be considered a better method to author clustering. Findings demonstrate that different co-author counting methods influence the network results of knowledge structure and a better counting method is revealed for author clustering.

초록

패스파인더 네트워크를 사용하여 지적 구조의 분석과 규명을 시도한 여러 연구가 발표되었다. 패스파인더 네트워크는 다차원척도법에 비해서 여러 장점을 가지고 있지만 구축 알고리즘의 복잡도가 매우 높아서 실행 시간이 오래 걸리며, 전통적인 지적 구조 분석에 유용하게 사용되어온 군집분석을 함께 적용하기가 어려운 것이 단점이다. 이 연구에서는 이와 같은 패스파인더 네트워크의 약점을 보완할 수 있는 새로운 기법으로 병렬 최근접 이웃 클러스터링(PNNC) 기법을 제안하였다. PNNC 기법의 클러스터링 성능을 전통적인 계층적 병합식 클러스터링 기법들과 비교해본 결과 효과성과 효율성 양면에서 기존 기법보다 우세한 것으로 확인되었다.

Abstract

Recently there are many bibliometric studies attempting to utilize Pathfinder networks(PFNets) for examining and analyzing the intellectual structure of a scholarly field. Pathfinder network scaling has many advantages over traditional multidimensional scaling, including its ability to represent local details as well as global intellectual structure. However there are some limitations in PFNets including very high time complexity. And Pathfinder network scaling cannot be combined with cluster analysis, which has been combined well with traditional multidimensional scaling method. In this paper, a new method named as Parallel Nearest Neighbor Clustering (PNNC) are proposed for complementing those weak points of PFNets. Comparing the clustering performance with traditional hierarchical agglomerative clustering methods shows that PNNC is not only a complement to PFNets but also a fast and powerful clustering method for organizing informations.