id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_fmyhnijisnd6vaii6at3k4sd3y	Jacques Savoy	Comparative evaluation of term selection functions for authorship attribution	2013.0	16	.pdf	application/pdf	10286	1214	72	To achieve this, we can consider different independent feature-scoring selection functions (information gain, gain ratio, term selection strategies have also been suggested in specific authorship attribution studies. In this study, we will consider words and punctuation symbols as possible features or terms. table, the value a indicates the number of texts belonging to the category cj in which the term tk occurs. To evaluate the different term selection functions, indicate that different selection functions may produce better performance levels than either the performance differences are usually statistically significant over the selection based on the most frequent words, but only for the two Sports corpora. Comparative evaluation of term selection functions Comparative evaluation of term selection functions Comparative evaluation of term selection functions Comparative evaluation of term selection functions Comparative evaluation of term selection functions Comparative evaluation of term selection functions Comparative evaluation of term selection functions	./cache/work_fmyhnijisnd6vaii6at3k4sd3y.pdf	./txt/work_fmyhnijisnd6vaii6at3k4sd3y.txt