| Popis: |
An analysis of text and image data is today one of the core fields of artificial intelligence. Among the means that can be used to process this information are structural methods. However, input chain deformations can often occur when analyzing data using structural methods. These are caused, for example, by the inaccurate recording of scanning means. In order to process inaccurate input information, it is necessary to extend a grammar describing the input objects to deformation rules and a weighing system indicating the degree of deformation of the rule. However, the expanded grammar is usually ambiguous. Specially designed syntax analyzers are required to process it. These analyzers may be time consuming; for example, Early Parser calculates a new state dynamically during the analysis. To accelerate processing, a decision table can be used, where every new state is pre-defined. To process distorted inputs, it is possible to use the modified Tomita parser, which contains mechanisms for processing ambiguities, while using the LR table, which reduces dynamically computed tasks. |
