A Hierarchical Statistical Framework for the Extraction of Semantically Related Words in Textual Documents