生醫研究資源群集判別

Community detection for Biomedical Research Resources

陳秋中(Chiu-Chong Chen)

指導教授 ： 陳建錦

國立臺灣大學/管理學院/資訊管理學研究所/碩士(2017年)

https://doi.org/10.6342/NTU201702368

摘要

本論文的主旨在於研究並利用智慧局部移動法(Smart Local Moving) 於生醫領域相關研究資源的群集尋找上，我們根據實際的平台資料先定義了一個網路模型，並利用點與點之間估計互信息(Estimating Mutual Information) 作為網路中節點間的邊之權重值。以這個模型我們建立了一個查找特定資源與最相關資源的服務。

此外根據這樣的模型我們利用SLM進行分群，並根據Dunn-Index與資料特性自定義了Coherence Index作為群集大小與數量的判別依據，最後以10-folds Cross-Validation進行實驗得出最佳的分群結果，供未來與相關領域專業人員，並進一步改良之討論依據。

關鍵字

網路分析 ； 學術網路 ； 群集辨識 ； 分群演算法 ； 群集驗證

並列摘要

A peculiar thing about biomedical researches is that they generally involve more than related past works and literature. Tools, softwares, databases and even samples are considered. Even though many online services have been developed to collect and share these resources. Still, they are too many for researchers to find desire information. For instance, SciCrunch, as one of largest online resource platform, contains more than 15,000 resources.

In this research, we present the analytic result for biomedical research resources in order to figure out meaningful groups of biomedical resources with community detection. We apply the Smart Local Moving algorithm to detect meaningful communities inside the network of resources.

並列關鍵字

Network Analysis ； Scholar Network ； Communities Detection ； Clustering ； Cluster Validation

參考文獻

1. Johnson, S. C. (1967). Hierarchical clustering schemes. Psychometrika, 32(3), 241-254.

2. Nascimento, M. A., Sander, J., & Pound, J. (2003). Analysis of SIGMOD's co-authorship graph. ACM Sigmod record, 32(3), 8-10.

4. Liu, X., Bollen, J., Nelson, M. L., & Van de Sompel, H. (2005). Co-authorship networks in the digital library research community. Information processing & management, 41(6), 1462-1480.

5. Tight, M. (2008). Higher education research as tribe, territory and/or community: A co-citation analysis. Higher Education, 55(5), 593-605.

6. Ding, Y. (2011). Scientific collaboration and endorsement: Network analysis of coauthorship and citation networks. Journal of informetrics, 5(1), 187-203.