Optimization of Transmission and Processing of Information of Electronic Documents Based on Parallel Computing Technology

A methodology has been developed to optimize the accuracy of information processing on conceptual principles of using statistical, natural, technological, semantic redundancy with mechanisms for extracting statistical, dynamic, specific characteristics, textural features and using probabilistic computing, searching, document recognition and clustering tools. A technique is proposed for monitoring and correcting errors of high multiplicity in textual information of documents based on morphological analysis and n-gram structured models. Tools for analyzing document information, performing arithmetic and logical operations of the same type, predictive analysis based on spline - functions, wavelet - analysis, and parallel computations have been developed. A technology has been implemented to optimize the reliability of electronic document information in the CUDA parallel computing environment based on the GPU model, mechanisms for extracting, matching frequently repeated word forms, derivational affixes and applying hash codes.

Перевод пока недоступен