My research interests combine formal syntax and semantics with dialogue and semantic parsing.
AMR/Meaning Representations for Situated Dialogue
- In collaboration with colleagues at the Army Research Lab (ARL), I am working on using Abstract Meaning Representation (AMR) as an interlingua for human-robot collaboration.
- Dialogue-AMR (Dial-AMR) presents an annotation scheme and corpus that captures speech acts, tense and aspect, and spatial relations.
- In collaboration with colleagues at Brandeis University, we have developed a continuation semantics for AMR and further annotated human-robot data with scope information to adapt AMR to this context.
Compositional Semantic Parsing Across Graphbanks
- In collaboration with a team at Saarland University, we are working on refining a compositional neural semantic parser that achieves competitive accuracies across a diverse range of graphbanks.
- Our team received 4th place overall, and 1st place on PSD, in the CoNLL 2019 MRP Shared Task.
- We continue to develop methods to normalize compositional structures across graphbanks for multi task learning and domain adaptation.
Multimodal Semantic Representation
- We’re organizing a workshop on multimodal semantic representations (MMSR), co-located with IWCS 2021!