AUTOMATIC EXTRACTING AS AN INTERACTIVE PROCESS

In our study we have tried to evaluate most of the existing methods of automatic extracting and to combine them into one single system for text reduction. Hitherto, all existing automatic extracting systems have produced only extracts of documents: they select certain sentences from the full text and list them in text order.

Despite the result that statistical methods can be used for extracting German language texts (and that similar methods show best results for German and English texts) the main outcome of this study was that computer-based methods for text extracting can alter the whole extracting and abstracting process as such and lead to completely new and as yet unknown aspects of text reduction: the tailormade extract and extracting as an interactive process.