Table 1.

Basic statistics of English corpus and samples

	0.1% Sample	0.5% Sample	1% Sample	Full corpus
N tweets (actual)	923,550	3,940,969	7,245,394	352,556,633
N tweets (rep.)	1,104,964	5,526,188	11,052,737	1,117,379,746
N tokens (actual)	28,754,912	109,303,100	199,369,396	9,134,879,457
N tokens (rep.)	31,236,676	156,348,389	312,746,780	31,292,640,403
Space savinga	16.42%	28.69%	34.45%	68.45%

^aSpace saving is calculated with the equation $1-\frac{\mathrm{Compressed}\mathrm{Size}}{\mathrm{Uncompressed}\mathrm{Size}}$, expressed as a percentage and using the number of tweets.