Scientometrics studies in quantitative fashion the evolution of science focusing on the analysis of publications.

One of its objectives is the development of information systems that can help to explore the enormous amount of scientific articles unceasingly published.

Our study proposes an information system to reconstruct a dynamical knowledge mapping from a functional data analysis perspective.

Starting database is a diachronic corpus which originates a words$\times$time-points contingency table displaying the frequencies of each keyword in the set of texts grouped by time-points in the observed time span.

The information system consists of an information retrieval procedure for keywords' selection and a two-stage functional clustering to reconstruct the historical evolution of the knowledge field under investigation.