BMBF funds Lucie Flek to establish an AI research group on Dynamically Social Discourse Analysis

DynSoDa Social Media User Representations BMBF

Prof. Dr. Lucie Flek receives a grant of over 1 Mil. Euro from the Federal Ministry of Education and Research (BMBF) for establishing an Independent Research Group on her project DynSoDA: Dynamically Social Natural Language Processing for Online Discourse Analysis. The 4-year project is a part of the BMBF support program for young researchers working in the field of Artificial Intelligence.

Lucie Flek points out that one of the challenges of today’s NLP models is that they typically assume one text has the same meaning for everyone. However, the message communicated is influenced by the individual characteristics of the individual and by the affinity to different socio-demographic groups. Current technological progress opens up a new opportunity to correct the decades-long deviation of machine learning from sociolinguistics, as deep neural models allow for a seamless integration of heterogeneous variables.

The DynSoDA project will model the discourse aspects of language together with the deep representations of user characteristics and latent social network profiles derived from online dialogues. In contrast to current approaches, user representations will be treated as dynamically contextual. The project further envisions the use of transfer and multitask learning techniques at multiple levels of abstraction to allow the framework to work robustly across a range of NLP tasks related to social discourse (such as opinion detection, sentiment analysis, hate speech identification, argument persuasiveness prediction, and rumor evaluation in social media), to better understand the similarities between these tasks, and to scale to numerous European languages. The innovative user representations, developed in this project, can also find future applications in the area of dialog modeling, improving quality and coherence of human-machine conversations.