Abstract
We propose to adapt an interval-based index structure, Relational Interval trees, to support the process of determining the structure of an unknown chemical compound. Important information for retrieving relevant substructures that make up a compound can only be described in an imprecise way, resulting in interval-based values specifying chemical shifts. The access method was implemented on top of a commercial database system and evaluated experimentally. The results of these experiments show that Relational Interval trees are an efficient way of indexing data needed for structure elucidation. © Springer-Verlag Berlin Heidelberg 2007.