Engineering the Production of Meta-Information: The Abstracting Concern

Abstract

In order to improve the automatic production of meta-information in the abstracting field, an essential starting point is the exposition of the current state of the art. At the level of content, three significantly different types of procedure stand out, depending on the document structure in question: extracting, rhetorical summarizing and cognitive summarizing. In addition, reticular and graphic models of information representation, much more appropriate to digital environments, offer a complementary method. In all cases, prior definition of the domain, with its specific documents and actors, is needed. However, the low quality of the product derived from full automation (extract and summaries), above all lacking in coherence, led us to the concept of partial automation, a hybrid man-machine methodology that, at least for the time being, seems to be the best solution for the abstract and abstracting problem.