A* CCG Parsing with a Supertag-factored Model

Abstract

We introduce a new CCG parsing model which is factored on lexical category assignments. Parsing is then simply a deterministic search for the most probable category sequence that supports a CCG derivation. The parser is extremely simple, with a tiny feature set, no POS tagger, and no statistical model of the derivation or dependencies. Formulating the model in this way allows a highly effective heuristic for A * parsing, which makes parsing extremely fast. Compared to the standard C&C CCG parser, our model is more accurate out-of-domain, is four times faster, has higher coverage, and is greatly simplified. We also show that using our parser improves the performance of a state-ofthe-art question answering system.

Broad-coverage CCG Semantic Parsing with AMR

“…We use the proxy report portion, which includes newswire articles from the English Gigaword corpus, and follow the official split for training, development and evaluation (6603/826/823 sentences). We use EasyCCG (Lewis and Steedman, 2014) trained with the re-banked CCGBank (Hockenmaier and Steedman, 2007;Honnibal et al, 2010) to generate CCGBank categories, the Illinois Named Entity Tagger (Ratinov and Roth, 2009) for NER, Stanford CoreNLP (Manning et al, 2014) for tokenization and part-of-speech tagging and UW SPF (Artzi and Zettlemoyer, 2013a) to develop our system. We use SMATCH to evaluate logical forms converted back to AMRs.…”

Joint A* CCG Parsing and Semantic Role Labelling

“…Then, after each item is added to the chart, the agenda is updated with all binary and unary rules that can be applied to the new item. For more details, see Lewis and Steedman (2014a).…”

“…If the chart size exceeds 20000 nodes, we back off to a pipeline model (roughly 3% of sentences). Finally, we build and use a tag dictionary in the same way as Lewis and Steedman (2014a).…”

“…This is particularly attractive for CCG, because the formalism's lexical and derivational ambiguity causes the parse charts to be very dense, in practice. Lewis and Steedman (2014a) showed that A * CCG parsing can be highly efficient, but used a restricted model without bi-lexical features.…”

“…Because CCG is strongly lexicalized, we are able to introduce a new type of extended lexical entries that allows us to factor the model over words and develop effective new upper bounds on the Viterbi outside parse score. A * parsing algorithms have previously been developed for models with tree-structured syntactic dependencies (Klein and Manning, 2003;Auli and Lopez, 2011b), and models with no bi-lexical dependencies, including supertag-factored CCGs (Lewis and Steedman, 2014a). We generalize these techniques to SRL-style graph-structured dependencies.…”