Formula Markup Algorithm for Search in Scientific Documents

This paper discusses the approach to a semantic search of mathematical expressions which allows seeking formulae by textual definitions of their variables. The proposed method allows establishing relations between terms, variables, and formulae. This method was tested in two different search services with different preprocessing approaches. The first service uses full-text indexing and responds to a query, the second uses ontology concepts and establishes relations in advance. In both cases, the proposed method produced high-relevant results. The method is applicable for search in documents containing mathematical expressions regardless the preliminary markup approach.