Jacob Andreas

3 paper cards in Mova's library · 2010–2019

Jacob Andreas wrote 3 of the papers Mova reads, in 2010–2019, most of them on Machine translation; often with Bevan Jones, Dan Klein, Daniel Bauer.

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.

kept — Error map: the workers failed mostly when the PP attached to a verb or an adjective, and rarely to a noun. Report PP head accuracy in en parse-eval split by the gold head's part of speech, on EWT test. (Towards Semi-Automated Annotation for Prepositional Phrase… · PP accuracy by gold head, EWT dev: noun 70.23 / verb 74.07 (missed = error), 79.08 / 84.46 among found; commit 35e51ff1b)

Pragmatically Informative Text Generation (2019)
Conditional generators often under-inform: a restaurant description may leave out attributes of its input meaning representation.
Semantics-Based Machine Translation with Hyperedge Replacement Grammars (2012)
The paper translates through a graph-shaped meaning representation. A source sentence is analysed into a graph, the graph is the pivot, and the target sentence is generated from it. The grammar is a weighted synchronous hyperedge replacement grammar. Its rules are learned from aligned string and graph pairs by an…
Towards Semi-Automated Annotation for Prepositional Phrase Attachment (2010) 1 idea kept
The paper asks whether untrained crowd workers can label PP attachment well enough to make training data. The system turns Penn Treebank parses into multiple-choice questions: the PP and its alternative noun or verb heads. The study used 1000 questions from Wall Street Journal sentences, each answered by three Amazon…