An Arabic Probabilistic Parser based on a Property Grammar

The specificities of the Arabic parsing such as agglutination, vocalization and relatively order-free words in sentences, remain a major issue to consider. To promote its robustness, parser should define different types constraints. Property Grammar formalism (PG) verify satisfiability constraints directly on units structure, thanks properties (or relations). In this context, we propose build probabilistic with syntactic properties, using PG, measure production rules terms implicit information particular properties. We experimented our treebank ATB algorithm CYK obtained encouraging results. Our method is also automatic for implementation most property type. Its generalization other languages or corpus domains (using treebanks) could be good perspective. combination pre-trained models BERT may make faster.