Zabolotnia T. The inverse context-associative method and software of automatized spelling correction of natural language objects

The inversion context-associative method of computer assisted correction of spellings errors is offered by means of iterative interleaving of procedures of hypotheses selection from a lexico-semantic dictionary on grounds of the content closeness to the context and their following check as for the correspondence to the formal criteria of similarity to the distorted word. The way of structural-algorithmic organization of spelling corrector in the form of specialized reactive software agent is given. It provides corrector’s interaction with the program modules of ATP open systems.The efficiency of the offered method of spelling correction and its software implementation according to the agent approach to the software development were studied experimentally on condition of their use in information retrieval system of Single information space of NTUU “KPI”. The got results confirmed theoretical positions of thesis work and allow to define the recommended option settings of algorithms of the errors correction, establishment of which provides effective work of the developed spelling corrector.