A hybrid clustering approach to identify network positions and roles through social network and multivariate analysis.

Autor/inn/en	Chang, Yu-Hsin; Lai, Kuei-Kuei; Lin, Chien-Yu; Su, Fang-Pei; Yang, Ming-Chung
Titel	A hybrid clustering approach to identify network positions and roles through social network and multivariate analysis.
Quelle	In: Scientometrics, (2017) 3, S.1733-1755 PDF als Volltext Verfügbarkeit
Dokumenttyp	gedruckt; online; Zeitschriftenaufsatz
ISSN	0138-9130
DOI	10.1007/s11192-017-2556-y
Schlagwörter	Role and position; Social network analysis; Multivariate analysis; Patent infringement litigation
Abstract	Abstract This study proposes a hybrid clustering approach to identify the positions and roles in a relational network by integrating multivariate and social network analysis. First, an adjacency matrix was constructed based on the graph theory to indicate the relation between the collected data. Next, network analysis was conducted and the statistics of network centrality as clustering variables were computed. After, this study reduced clustering variables using the principal component analysis. These selected principal components were then used as clustering variables for a two-step cluster analysis. Hierarchical cluster analysis was first made to determine the appropriate number of clusters and then K-means clustering was used for dividing actors into k proper positions. In addition, the multivariate analysis of variance was conducted to test the significance between those positions. After, a new adjacency matrix was built upon the rearrangement of k positions. The frequency within and between these positions was computed and the cut-off value was determined to distinguish the difference between these frequencies. Finally, each position was labeled based on its characteristics and the relationships within and between these positions. After the structured approach was established, the litigation-related network of smartphone makers was used as empirical evidence. The results showed that this structured approach can effectively distinguish the position and role of a company in a relational network.
Erfasst von	OLC
Update	2023/2/05