Bisecting Tensor Decomposition to Discover Theme Changes over Time

摘要：

With the goal of helping archivists analyze document collections we developed a bisecting tensor decomposition method to detect theme changes and relationships across time and provenance automatically. The results show that the method highlights relevant features in the collection and traces their evolution over time. This is consistent with the way in which archivists describe collections for access purposes.

关键词： tensor analysis digital archives text mining

作者: Pradeep DHANANJAY Weijia XU Maria ESTEVA Victor EIJKHOUT

作者单位: Texas Advanced Computing Center The University of Texas at Austin, Austin. Texas, USA

会议类型: 国际会议

会议名称: International Council for Scientific and Technical Information Annual Conference(国际科技信息委员会2011年夏季年会 ICSTI 2011)

会议地点: 北京

会议语种:英文

页码: 176-180

在线出版日期: 2011-06-07（万方平台首次上网日期，不代表论文的发表时间）