Automatic Summarization of Text Messages in Group Chats Based on Parent Chats on WhatsApp

Automatic text summarization is one of the key tasks of NLP, which consists in creating a short version of the source text. In the modern world the volume of information consumed by people is constantly growing, therefore the task of summarization is receiving more and more attention. Despite significant advances in the summarization of scientific and news articles, the methods applied to monologue texts are not always effective in dialogue summarization. This paper is devoted to fine-tuning models for the task of abstractive summarization for the Russian language on a custom dialogue dataset. A parental chat with a teacher in WhatsApp was used as material for the dataset. The ruT5, mT5 and RuGPT models were selected for fine-tuning, and the ROUGE–1, ROUGE-2, ROUGE-L, BLEU and BERTScore metrics were used to assess their quality. As a result, the ruT5 model, fine-tuned on the custom dataset, managed to outperform the baseline model in four out of five metrics, second only in the BLEU value.