A study on statistical generation of a hierarchical structure of topic-information for multi-documents
Viet-Cuong Nguyen · Institutional Repositories DataBase (IRDB) · 2011
Generating a hierarchical structure of topic-information (HST) for multiple documents written about the same topic is a new task in natural language processing.In a HST, topic-information is represented in a phrase and it can be seen as a title.Intuitively, a HST looks like a table-of-contents which is normally presented at the beginning of a book.It could play as a navigation tool to help readers quickly locate interesting parts.In addition, readers could look through a HST to get an overview of the topic of the document set.In this study, we propose a framework for generating a HST for multidocuments which involves three sequential tasks: