Mike Lewis

5 paper cards in Mova's library · 2014–2017

Mike Lewis wrote 5 of the papers Mova reads, in 2014–2017, most of them on Semantic role labeling, Coreference resolution; often with Luke Zettlemoyer, Kenton Lee, Luheng He.

The profile is the library's match of the name; two people with one name are not always told apart.

Papers

End-to-end Neural Coreference Resolution (2017)
The first coreference model trained end to end without a parser or a mention detector. Every span up to 10 words is a candidate; each span gets a mention score and each pair an antecedent score, and a span chooses among all earlier spans plus a dummy antecedent fixed at score 0 (no link, or not a mention). Span…
Deep Semantic Role Labeling: What Works and What's Next (2017)
A neural SRL model that treats each predicate's arguments as BIO tags over the sentence, with no syntax input.
LSTM CCG Parsing (2016)
The paper asks whether a CCG parser needs an explicit model of bilexical dependencies. Its parser has none: a bidirectional LSTM supertagger assigns lexical categories, a deterministic grammar combines them, and A search finds the best parse. Semi-supervised tri-training adds training signal. On CCGbank section 23…
Question-Answer Driven Semantic Role Labeling: Using Natural Language to Annotate Natural Language (2015)
The paper introduces question-answer driven semantic role labeling (QA-SRL): each verbal predicate is labeled with wh-questions, such as "Who finished something?", and each question is answered by phrases of the sentence.
A CCG Parsing with a Supertag-factored Model (2014)
The question is whether CCG parsing needs a full model of derivations and dependencies, or whether a model of lexical categories is enough.