The use of additional evidence in mining usercreated descriptions for content structural design

Abstract

The use of a text mining approach for full automatic taxonomy creation for content management has proven with serious limitations. The high level semantics indicating relevant association of entities among the documents are often not explored. This study introduces a feasible method that allows identifying high level semantics into text mining procedures while providing for appropriate levels of document descriptions to support access and discoverability. Due to the effectiveness of categorization and adequacy of the structure created can be better determined by humans who are familiar to the documents, qualitative inquiry rather than a purely experimental design was applied. The study collected the data and run the text mining analysis with text analysis, clustering and topic extraction. Two examples show how to develop a faceted classification structure to support digital collection access and navigation using the method. The study indicates that the text-mining method supports taxonomy creation with more efficiency and accuracy when human domain and application knowledge are captured during data collection and text mining processing. The proposed method of taxonomy creation would support the creation of new knowledge.