SocialBERT – Transformers for Online Social Network Language Modelling

P. 1–10.

Ilia Karpov, Nick Kartashev

The ubiquity of the contemporary language understanding tasks gives relevance to the development of generalized, yet highly efficient models that utilize all knowledge, provided by the data source. In this work, we present SocialBERT - the first model that uses knowledge about the author’s position in the network during text analysis. We investigate possible models for learning social network information and successfully inject it into the baseline BERT model. The evaluation shows that embedding this information maintains a good generalization, with an increase in the quality of the probabilistic model for the given author up to 7.5%. The proposed model has been trained on the majority of groups for the chosen social network and is still able to work with previously unknown groups. The obtained model, as well as the code of our experiments, is available for download and use in applied tasks

Keywords: natural language processing, social network analysis, deep learning, Neural Language Processing (NLP), BERT, BERT, Language Modelling, Knowledge Injection, Language Modeling, Knowledge Injection

Publication based on the results of: Applied Network Analysis for the Solution of Current Problems of the State, Business and Society: Methodological Developments and Practical Implementations (2022)

In book

Analysis of Images, Social Networks and Texts. 10th International Conference, AIST 2021, Tbilisi, Georgia, December 16–18, 2021, Revised Selected Papers

Cham: Springer, 2022.