2T: Two-Term Indexing of Documents Using Syntactic and Semantic Constraints
Harri M. T. Saarikoski · 2006
Purpose of an index is to provide an intuitive navigational structure for the user into the subject matter. This paper specifies a novel, untrained method (2T) for automatically producing a two-level, semi-formal, concept-based index out of textual documents, which consists of topics rather than keywords. Using syntactic and semantic constraints and a domain ontology containing the relevant terms, we obtain high accuracy (high 80's to low 90's). Using concept categories (or semantic roles) to validate sensicality of an index term is a novel approach in semantic indexing. Resulting low-cost navigational structure can add value to businesses relying on accurate automatic document indexing (e.g. mobile news providers) or dealing with a critical learn-to-do requirement for their employees or clients (e.g. airplane maintenance). It can be implemented as a post-processing stage of full-text indexes or as readable and editable index to documentation - either printed or online.