NON-STATISTICAL METHODS OF AUTOMATIC EXTRACTION OF CAUSAL RELATIONSHIPS FROM THE TEXT

Most of the first attempts to extraction causal relationship were tied with complex and manuallinguistic patterns, syntactic rules small datasets based on domain. This article examines paradigmof a non-statistical approach relationships, its basis, language constructs,patterns, classification relationships. The aim was study methods thisparadigm, determine their disadvantages, advantages, possibility application.The discusses various approaches given by authors well-known highly cited researchpapers impact success analysis thesescientific papers has unequivocally confirmed that task extracting CR is an extremely difficult taskof natural processing. presence variety linguistic constructions language,ambiguities kinds, as well features greatly affect accuracy extraction.Almost all have encountered problem specialized fields ofknowledge, where expert description almost always required. Also, methodsare manual or semi-automatic, because assume construction templates for determining inthe text. Even though non-static sufficient (on average 70-80%) successfullycope under consideration, there currently no universal method CR.The proposed should be respect languages, subjectareas defining implicit CR.