Low-level structural recognition of documents
Yannick Chenevoy, A. Belaied · OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) · 1994
This paper focuses on the qualitative approach of the low-level structured document analysis. The system identifies the different logical fields within the document and produces as output a structured flow with confidence scores. The strategy is driven by a generic model and by an OCR flow. Logical labels are attached to research areas after hypothesizing and testing typographical, lexical and contextual properties. A qualitative recognition is performed, which allows to amphasize ambiguities and unrecognized fields. Library references are treated to illustrate this method.