Recognition of Situations Described in the Text of Legal Documents

The article investigates the use of machine learning in solving the problem of natural language processing - recognition of situations described in the legal document. The purpose of situation recognition is establishing a procedure that is then to be performed. This task can be referred to as a classification task. Therefore, for its solution the neural networks can suit very well. To understand how to recognize a situation, the recognition process is built on the basis of the situational approach. In the situational semantics conclusions are made only within the situation described in the text. The situation may be described by one or more sentences. In the situational approach each situation is assumed to correspond to a finite number of different discrete one-step control actions. One-step actions make up the procedure. The nuclear semantic structure is used to describe the situation. This approach to the description of situations and identification of appropriate procedures allows formalizing the description of the situation, and then forming a training set. The research on the use of neural networks to solve this task has yielded positive results.