This paper presents a natural language processing semantic modeling approach that can be automated to pull conceptual information from text documents. The approach is based in the use of theta-roles, in particular thematic-hierarchies, to create a collection of vectors culled from the sentences of a document that describe content of the document. This approach can form the basis of a natural language processing-based concept representation scheme.
Carlisle, Judith, "Semantic modeling usuing theta-roles: a natural language processing-based approach to conceptual representation" (2010). AMCIS 2010 Proceedings. 437.