Our main research contributions in the field of programming languages are: -Invention of multiple attribute grammar inheritance that enables incremental programming language development. -We have shown how other language-based tools, not only compilers but also visualizers and animators, can be automatically generated from formal language definition. -Existing patterns for domain-specific language development have been extended and newly classified. -Design and implementation of context-free grammar induction algorithm based on evolutionary computation using positive and negative samples only. Each research activity from the area of computational geometry resulted in an original solution.Either a new method, more efficient of known methods or at least comparable to them, or an acceleration of some known algorithm have been developed. Let us emphasize the following solutions: -Original acceleration technique of two-level uniform space subdivision was built in many algorithms of computational geometry. It gives excellent results when the input data are distributed distinctly nonuniformly. -New sweep-line-based algorithm of Delaunay triangulation is significantly faster than the existing methods of Delaunay triangulation. -Our algorithm for compression of triangular meshes currently represents the most efficient solution for this problem. -Design and implementation of an efficient algorithm for creation of geometric buffers. -New sorting algorithm ovrcomes drawbacks of quicksort in case when the input data are already completely or nearly sorted. -Algorithm of spatial data visualization "differential splatting" efficiently renders both, statical (3D) and dynamically changeable (4D) spatial data. A new method for surface flattening based on divide and conquer strategy has been developed. In the field of procedural textures, a lot of new methods have been constructed. Current methods for cloud generation have been simplified and quickened. A new algorithm for blending implicit primitives that correctly arranges cloud density has been developed. As first we have used hypertextures for tree modeling.A new algorithm for curve reconstruction based on Euclidean minimal spanning tree has been made. Our research in the field of electromyography (EMG) resulted in two novel algorithms for decomposition of compound signals.One is based on hihger-order cumulants, while the other utilizes blind source separation.Both approaches were sucessfully applied to the surface EMG analyses, leading to some new, clinically significant parameteres.These algorithms bring novelty in a world-wide sense, in particular their implementation in the surface EMG decomposition. With these solutions we participated in the NEW project, conducted within European 5th Framework. Detection and modelling of human organs based on ultrasound and magnetic resonance imaging also generated new approaches.Using image sequences, we are able to reconstruct ovarian follicles from ultrasound recordings and knee-joint structures from magnetic resonance recordings.The former enhances the so called "in vitro" fertilisation, while the latter simulates the pre-and postoperative condition of an injured organ. The knee dynamic simulations were the main topics in the European 5th Framework project entitled SimBio. Our most important contribution in the field of conceptual learning/teaching tools is a simulator of postnatal complications.It was implemented in the form of virtual delivery room and virtual baby, which introduces a new concept of virtual medical training. The prototype implementation is installed at Stanford University, California, USA. We have developped a framework for building portals which cosists of horizontal services.These services are authentication service, integration service, personalization service, notification service, and user interface service.We achieve important research results on building inteligent user interface services.