id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_n3svrctkqnaoxhgk3g43mf6bly	Maciej Eder	Does size matter? Authorship attribution, small samples, big problem	2013.0	4	.pdf	application/pdf	1762	166	65	size of text samples for authorship attribution different sample lengths, languages and genres acceptable sample length for reliable attribution word frequency strongly depends on the size further increase of input sample size would not affect the effectiveness of the attribution. corpus, 500 randomly chosen single words were samples were analyzed using the classical Delta samples from the original texts, followed by the 20000 words per sample. The results for a corpus of 63 English novels which indicates the minimal sample size for the that samples shorter than 5000 words provide a 3000 words, the obtained results are simply sample (and there was no significant difference corpora: English and Latin (3500 words per sample were enough for good results), and, minimal effective sample size was of some 2500 was to test the attribution effectiveness of this depends to some extent on the word-chunk size acceptable sample size for future attribution Attribution Studies: Some Problems and	./cache/work_n3svrctkqnaoxhgk3g43mf6bly.pdf	./txt/work_n3svrctkqnaoxhgk3g43mf6bly.txt