id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_okkzldu7ozgi5bflm7hthf7sgq	Damiano Spina	Discovering filter keywords for company name disambiguation in twitter	2013	38	.pdf	application/pdf	16299	1657	68	keywords cover, in average, 28% of the tweets for a company in our test collection. Keywords: Twitter, online reputation management, name disambiguation, filtering terms extracted from tweets retrieved by querying the company name in Twitter. A positive/negative filter keyword is an expression that, if present in a tweet, indicates a manual selection of positive and negative keywords for all the companies in the WePS-3 Table 1: Differences between oracle and manual positive keywords for some of the company names on the test collection. Table 2: Differences between oracle and manual negative keywords for some of the company names on the test collection. 1. Collection-based features (col * ): Terms that co-occur frequently with the (ambiguous) company name c, or terms written as hashtags should have more probability to be (positive/negative) keywords than others. the set of tweets Tc (8), where fti is the web-based feature value f for the term ti	./cache/work_okkzldu7ozgi5bflm7hthf7sgq.pdf	./txt/work_okkzldu7ozgi5bflm7hthf7sgq.txt