Faceted Navigation on Text

This paper proposes a method which is not for summarization but for extracting multiple facets from a text according to the keyword sets representing readers´ interests, so that readers can obtain the interested facets and carry out faceted navigation on text. A facet is a meaningful combination of the subsets of the text. Previous text process technologies are mostly based on text features such as word frequency, sentence location, syntax analysis and discourse analysis. These approaches neglect the cognition process of human reading. The proposed method considers human reading process. Experiments show that the facet extraction is effective and robust.