A hierarchical and recursive model of mathematical expressions for automatic reading of mathematical documents

We propose an original model for automatic reading of mathematical texts, founded on a recursive approach and a hierarchical view of the latter. This hierarchy in the text has been introduced by means of the notion of mathematical object, that we characterise by a set of definitions. A prior segmentation phase of mathematical objects in text is performed, using character matching techniques and propagating the labelling process of mathematical objects around “special characters” initially recognised. Complex and dynamical tree representations are proposed for mathematical expressions detected in the text, and a dynamical set of priority rules has been defined that guides this automatic process. Finally the recognition of the elementary objects obtained permits to achieve the encoding of a mathematical expression