Recognizing Documents versus Meta-Documents by Tree Kernel Learning

The problem of classifying text with respect to metalanguage and language object patterns is formulated and its application areas are proposed. Examples of metalanguage patterns in text are foreign language grammar lessons and tutorials on how to write engineering documents. The method targets the text classification tasks where keyword statistics is insufficient do distinguish between such abstract classes of text as metalanguage and object-level. To do that, we extend the parse tree kernel method from the level of individual sentences towards the level of paragraphs. We build a set of extended trees for a paragraph of text merging individual parse trees for sentences. We evaluate our approach in the domain of the design documents, differentiating them from meta-documents such as instructions on how to write design documents.