Applying software analysis technology to lightweight semantic markup of document text

Software analysis techniques, and in particular software “design recovery”, have been highly successful at both technical and business level semantic markup of large scale software systems written in a wide variety of programming languages, and in particular have proven efficient and scalable in assisting the resolution of the “year 2000” problem for billions of lines of legacy source code. In this work we describe a first experiment in applying the same technical solutions and tools that have proven so successful in software markup to the more general problem of semantic markup of text documents. In this early report we describe our adaptation of the software analysis techniques, propose a general domain-independent architecture for semantic markup using them, and demonstrate its feasibility in a limited but realistic domain of application by comparison with both raw and tool-assisted human semantic markers.