Transformations of vertex sequences of regular grid graph into paths of an arbitrary connected graph are facilitated according to various coarsening and approximation operations, including minimum cost alterations and minimum cost re-routings. The sequence transformations are supposed to support issues of man-machine interaction, which implies lack of an ultimate formal design objective. Furthermore, this implies that formal methods and algorithms have to be combined in a pragmatic fashion. For planar graph, the notion of Voronoi regions is modified to graph Voronoi regions which partition the plane according to proximity to verttices and edges simultaneously. The non-planar case is reduced to the planar case by adding all intersection points of vertex connections to the original vertex set and by splitting vertex connections accordingly. This allows grid point sequences to be intermediately transformed to so-called mixed or region sequences which are eventualy transformed to vertex sequences by production rule-like operations. The algorithmic preprocessing burden of partitioning and indexing the euclidean plane via the graph Voronoi regions or approximations thereof is practically larger and typically more complicated than any of the run time computations.
The paper presents the idea of using regular expressions in translation systems based on translation memories. Regular expressions (regexps) are used for: search for a match in the input sentence, search for an appropriate example in the translation memory and in the transfer of the input sentence into its equivalent. The application of transfer rules to translation memories supports the thesis, put forward in the paper, that Machine Translation and Computer-Aided Translation converge into the same direction.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.