id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_snkm6xv4wzaolcpximorxj4j2m	Mike Lewis	Improved CCG Parsing with Semi-supervised Supertagging	2014	12	.pdf	application/pdf	6918	687	59	Although CCG parsers perform at state-of-the-art levels (Rimell et al., 2009; Nivre et al., 2010), fullsentence accuracy is just 25.6% on Wikipedia text, unlikely to be made available, recent work has explored using unlabelled data to improve parser lexicons (Thomforde and Steedman, 2011; Deoskar et similar techniques to CCG supertagging, hypothesising that words which are close in the embedding Recent work has explored using vector space embeddings for words as features in supervised models for a variety of tasks, such as POS-tagging, We introduce models for predicting CCG lexical categories based on vector-space embeddings. Dimensionality is the set of dimensions of the word embedding space that we experimented with, and Training Words refers to the size of the unlabelled corpus the other embeddings would perform better with different training data, dimensionality, or model architectures. Table 3: Comparison of different model architectures, using the Turian embeddings and a 5-word	./cache/work_snkm6xv4wzaolcpximorxj4j2m.pdf	./txt/work_snkm6xv4wzaolcpximorxj4j2m.txt