Among the solutions for the detection of cause-and-effect relationships are methods based on knowledge, statistical solutions or methods allowing the use of deep learning. The solution presented in the article uses bidirectional artificial neural networks LSTM to detect such relationships in legal texts in Polish. The analysis was performed at the sentence level, but due to the specific legal language and the focus on Polish, two separated networks were used in the experiment. The task of the first one is to classify whether a sentence contains a~conditional, while the second one is to identify the elements of this relationship. Both use word embedding sets for the Polish language corpus. The results of the experiment prove that it is possible to perform such extraction with satisfactory results, and raise questions and point to further possible ways forward.
JavaScript jest wyłączony w Twojej przeglądarce internetowej. Włącz go, a następnie odśwież stronę, aby móc w pełni z niej korzystać.