Vincent Ng

4 paper cards in Mova's library · 2010–2022

Vincent Ng wrote 4 of the papers Mova reads, in 2010–2022, most of them on Coreference resolution, Discourse analysis; often with Michael Strube, Carolyn Rosé, Juntao Yu.

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

Ideas Mova took (1)

From the ideas ledger: an idea tried, kept or queued, or measured with no gain, and what it gave.

measured, no gain — Implement the discourse-deixis baseline as a sieve in coref/src/sieve.rs (the disabled demonstratives slot is the place): this/that (and it as subject of happened, was true, means) not followed by a noun → the preceding clause, as an event mention with a… (The CODI-CRAC 2022 Shared Task on Anaphora, Bridging, and… · deixis sieve (event «it», bare this/that → previous clause, as an added clause mention): 17 of 141 right on GUM dev, −0.76/−0.09 CoNLL gold; in coref, off by…)

The CODI-CRAC 2022 Shared Task on Anaphora, Bridging, and Discourse Deixis in Dialogue (2022)
The second edition of a shared task on anaphora in dialogue, with three tasks: identity anaphora (including split antecedents and singletons), bridging reference, and discourse deixis (a pronoun such as that or this referring to an event, fact or utterance).
The CODI-CRAC 2021 Shared Task on Anaphora, Bridging, and Discourse Deixis in Dialogue (2021)
The shared task asks systems to find anaphoric relations in conversation, a genre that OntoNotes-style corpora barely cover.
Scoring Coreference Partitions of Predicted Mentions: A Reference Implementation (2014) measured
Coreference metrics such as MUC, B-cubed and CEAF were originally defined for gold mentions, and different research groups adapted them to predicted mentions in incompatible ways, often by adding or removing mentions artificially so that the key and response partitions line up.
Supervised Noun Phrase Coreference Research: The First Fifteen Years (2010)
A survey of supervised noun-phrase coreference research over about fifteen years. It follows three model families: the mention-pair model (with its instance creation, training and clustering steps such as closest-first and best-first clustering), entity-mention models that use cluster-level features, and ranking…