Automatic Сlassification of Narrative Structure of Texts of Personal Diaries

The task of automatic determination of the time reference of event is one of the central objects of interest in the field of information retrieval and facts extraction. This research is devoted to solving the problem of automatically determining the time reference of event on the material of personal diaries. To reach this goal narrative structure of diaries was studied. Extensive corpus of diaries texts was collected and a system of classes clause was developed, which allows to provide the necessary types of clauses for further study. A system of parameters was established and an algorithm of marking these parameters was created. With a help of this algorithm, the corpus of diaries texts was marked. Educational corpus was manually collected and marked, which was used for machine learning algorithms. The classification of the diary was produced, which allows to identify clauses suitable for the further extraction of event-related entities and temporal references.