Scholar
Hannah Rashkin
Google Scholar ID: gFwDGB4AAAAJ
Google Deepmind
Citations & Impact
Citations
11,019
H-index
20
i10-index
20
Publications
20
Co-authors
0
Contact
Publications
Resume (English only)
Academic Achievements
- Attribution Metrics: Human evals of attribution in grounded model output
- The Begin Dataset: For meta-evaluation of attribution metrics on knowledge-grounded dialogue
- Navigating tradeoffs between attribution and specificity in dialogue
- Increasing Faithfulness with Control Codes
- Social IQA Dataset: New benchmark for measuring models' ability to reason about social interactions
- Grover: Defending against Fake News
- Story Commonsense Dataset: Inferring changes in character motivations and emotional reactions in short commonsense narratives
- Multilingual Connotation: Multilingual analysis of connotation over twitter
- EmpatheticDialogues Dataset: New dataset and task for empathetic response generation
- ATOMIC: Large-scale commonsense knowledge graph for if-then reasoning about events
- Event2Mind: Pragmatic inference of the intent and reactions of participants in events
- Connotation Frames: Investigating predicate-specific lexicons for connotative relationships implied by word choice
- COMET: Commonsense Transformers for knowledge graph construction
- Power & Agency in Movies: Incorporating power and agency to connotation frame lexicon with analysis of gender bias in movies
- Fact-Checking: Linguistic analysis of subtle persuasive techniques to detect truth-bending
- Document-level Sentiment Inference with Social, Faction, and Discourse Context: Incorporating social theories to perform entity-entity sentiment inference at a document level
Research Experience
- Worked as an intern at MSR (summer 2019), a part-time research intern at the Allen Institute of Artificial Intelligence (2018-2019), an intern at FAIR (summer 2018), and an intern at Pacific Northwest National Laboratory (summer 2016).
Education
- Received a PhD from the University of Washington in June 2020, advised by Prof. Yejin Choi. Undergraduate student at the University of Illinois at Urbana-Champaign.
Background
- Currently a research scientist at Google Deepmind. Interested in designing and evaluating NLP systems that are more knowledge-grounded, controllable, interactive, and capable of complex reasoning. Also interested in positive applications of NLP models for real-world challenges and the intersection of NLP and computational social science.
Co-authors