CNPq announced the approval of funding for the project “Semantic Representation of Multimodal Objects: bringing Computer Vision and Natural Language Understanding together via Frame Semantics“. The proposal is a joint initiative of FrameNet Brasil | UFJF and LETra | UFMG laboratories exploring how FrameNet can be used to build a gold standard corpus annotated for frames and for both language and video modes. The project will also build a model for tagging audiovisual objects for semantic frames.
Although excited about the agency’s recognition of the innovative and audacious nature of the proposal, Dr. Torrent noted the difficulties faced at the moment: “Unfortunately, the project budget suffered a severe cut due to the financial restrictions imposed on the National Research Council by the Brazilian Government. We’ll keep our pursuit for additional sources of funding and our fight for more investment in science in Brazil”.