Method and device for recognising and classifying sections of a document which can be accessed on a computer by means of step-by-step learning during training sessions

摘要

The method involves partitioning a computer-accessible document into document parts. Types e.g. text types, of the document parts are determined by a start-classifying factor for each part. A classification-security measure by which the types are determined, is calculated by the factor for the document. A specification of the types is detected for the parts when the measure is smaller than a preset limit. A classification factor is generated for the document of training sequences. The start-classification factor is trained with the detected types of the document parts during the generation. Independent claims are also included for the following: (1) a data processing system for generating a classification factor for analyzing a computer-accessible document (2) a data carrier comprising instructions for performing a method for analyzing a computer-accessible document.