It is impossible to imagine modern scientific activity without the use of intelligent information systems. Modern scientific information systems include electronic libraries, scientific project management systems, smart search engines, scientific information social networks, etc. In this article, we reviewed the most popular and widely used electronic libraries today, as well as some other scientific information systems. A number of shortcomings in the organization of classical electronic libraries based on relational databases have been identified: the relational schema is rigid and its change leads to the need to change a large amount of program code; on the basis of relational schemas it is impossible to draw any logical conclusions; description of the content using the code of the system of universal decimal classification of information and a set of terms has very limited possibilities for assessing the semantic proximity between objects of the same type and, moreover, of different types; the use of relational schemas significantly limits the interaction between different electronic libraries. As an alternative model of electronic libraries organization, ontological models are considered, which are considered more perfect and expressive. They solve the problems of describing the semantics (meaning) of the content of electronic libraries and implementing work with them using such knowledge representation models as semantic networks, frames and ontologies. The analysis of the problems and prospects of intellectual information systems in science has been carried out.