TYPO CORPUS

What's new

• 2010-01-17: This page is available.

List of Contents

• Corpus
• What's new
• Download

Corpus

This corpus is a collection of typos in tweets.
The corpus consists of pairs of a typo and its original form (such as gogle:google).
The typos dealt with this research are classified into four types:
• (1) INSERT (IN): a character is added to the original word.
• (2) REMOVE (RM): a character is removed from the original word.
• (3) REPLACE1 (R1): the order of character is different from the original word (the number of differences is one).
• (4) REPLACE2 (R2): a character is different from the original word.

Format

typo	original	operation	operation symbol*	context of typo (freq)	context of original (freq)
intvite	invite	IN	in < t > vite	google_wave_intvite(2)	google_wave_invite(38802)
goole	google	RM	goo(g)le	my_goole_wave(1)	my_google_wave(35841)
goolge	google	R1	goo[l/g]e	a_goolge_wave(1)	a_google_wave(42205)
waze	wave	R2	wa[z:v]e	google_waze_invite(2)	google_wave_invite(38802)

* the operation symbol represents the operation. < > indicates IN; () indicates RM. [/] indicates R1, and [:] indicates R2.

Download

• typo-corpus-r1.txt