Topical classification of text fragments accounting for their nearest context

We present an approach to topical classification of biographical text fragments that takes into account the nearest context of classified fragments using a neural network with several inputs. The choice of the model architecture is based on the assumption that since texts written in a natural language differ in consistency and coherence, the context of a passage can be used as additional input data. The model was trained and tested on the biographical corpus compiled by ourselves. The results obtained using the proposed approach outperformed the results of models that do not take into account the context of the passage.