On Semantic Search Algorithm Optimization

Abstract

In the article we consider, on the example of development of a relational database (RDB) information system for Tatneft oil and gas company, an approach to organization of effective search in large arrays of heterogeneous data, satisfying the following essential requirements. On the one hand, the data is integrated at the semantic level, i.e. the system supports the presentation of data, describing its semantic properties within an unified subject domain ontology. Accordingly, end user's request are formulated exclusively in the subject domain terminology. On the other hand, the system generates unregulated SQL-queries, i.e. the full text of possible SQL-queries, not just values of particular parameters, predefined by the system developers. Considered approach includes both the possibilities of increasing the reactivity of the universal SQL queries generation scheme as well as more specific optimization possibilities, arising from the particular system usage context.