id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_aawrwejmaregji2stj32j4azfq	Dingquan Wang	Fine-Grained Prediction of Syntactic Typology: Discovering Latent Structure with Supervised Learning	2017	16	.pdf	application/pdf	10033	1300	71	as supervised learning, using a large collection of realistic synthetic languages as training data. (hand-engineered or neural features) that correlate with the language's deeper structure (latent trees). Table 1: Three typological properties in the World Atlas of Language Structures (Dryer and Haspelmath, 2013), and how they the non-IID problem that the available OVS languages may be evolutionarily related.1 We mitigate this issue by training on the Galactic Dependencies treebanks (Wang and Eisner, 2016), a collection of more than 50,000 human-like synthetic 1Properties shared within an OVS language family may appear to be consistently predictive of OVS, but are actually confounds that will not generalize to other families in test data. To score all dependency relation types given the corpus u, we use a feed-forward neural network with the model to predict p∗train from utrain on each training language. Table 3: Average expected loss over 20 UD languages, computed by 5-fold cross-validation. of relation directionality for each training language.	./cache/work_aawrwejmaregji2stj32j4azfq.pdf	./txt/work_aawrwejmaregji2stj32j4azfq.txt