Learning-based Transformation for Text Documents

摘要

This paper presents a method to automatically transform semistructured (not necessarily tagged) text documents into content-tagged documents based on techniques from machine learning and computational linguistics. The method consists of two phases. First, a learning-based segmentation module is used to extract regions and sequences from the documents. Second, translation from region-marked documents to XML is done by a transformation-based learning (TBL) translator that is very effective even with a small set of training examples.