Towards a Comparison of Structured and Unstructured Text Retrieval

Abstract

Structured text retrieval is the possibility of using the structure of documents (e.g., SGML tags) in search criteria, in addition to full-text operations (adjacency, etc.). Intuitively, using the structure can improve the performance of the retrieval, but for many reasons, it is difficult to compare structured and unstructured retrieval by real experimentation, and it has not been done yet.