id	author	title	date	pages	extension	mime	words	sentence	flesch	summary	cache	txt
dad-11372	Zeldes, Amir; Liu, Yang	A Neural Approach to Discourse Relation Signal Detection	2020	33	.pdf	application/pdf	16391	583	45	Amanda Stent Submitted 09/2019; Accepted 02/2020; Published online 07/2020 Abstract Previous data-driven work investigating the types and distributions of discourse relation signals, including discourse markers such as ‘however’ or phrases such as ‘as a result’ has focused on the relative frequencies of signal words within and outside text from each discourse relation. In fact, only about 8.3% of the tokens correctly identified by the model in Table 6 below are of the DM type, whereas about 7.2% of all tokens flagged by human annotators were DMs, meaning that the model frequently matches non-DM items to discourse relation signals (see Performance on Signal Types below).	cache/dad-11372.pdf	txt/dad-11372.txt