WP5: Hybrid AI for Linguistically-Informed Processing of Texts & Conversations
Joty Shafiq Rayhan (Nanyang Technological University)
WP5 focuses on developing applications able to extract deep semantic and pragmatic information that hinges on structural properties of discourse and conversations by (1) proposing advanced knowledge-enhanced deep learning models for natural language understanding that inject and maintain appropriate linguistic knowledge, in particular the right sort of linguistic structures that will enable us to extract desired information, and (2) incorporating the human-in-the-loop approach by maximizing models social impacts and enforcing their acceptability. Both objectives have to do with the data bottleneck. The project will propose solutions to overcome the lack of large amounts of data to train deep learning models, such as generating training data from noisy sources through hybrid AI. We illustrate our approach with four main themes that we believe are of particular importance for AI-based urban systems: (1) interpretative bias for transparent and trustable AI, (2) contextual sentiments/emotions, (3) detecting user intentions and (4) modelling of conversations. WP5 will interact with WP1 (strategies for models explicability to enforce their acceptability), WP6 and WP7 (public feedback over social media) and WP2 (data augmentation techniques).
This research is supported by the National Research Foundation, Prime Minister’s Office, Singapore under its Campus for Research Excellence and Technological Enterprise (CREATE) programme.
CREATE is an international collaboratory housing research centres set up by top universities. At CREATE, researchers from diverse disciplines and backgrounds work closely together to perform cutting-edge research in strategic areas of interest, for translation into practical applications leading to positive economic and societal outcomes for Singapore. The interdisciplinary research centres at CREATE focus on four areas of interdisciplinary thematic areas of research, namely human systems, energy systems, environmental systems and urban systems. More information on the CREATE programme can be obtained from www.create.edu.sg.
Visit the CNRS website here.