IPSJ SIG Technical Report 2009-CSEC-46 No.38
ipsjstyle-ms2009.dot (V2.0)
CAPTCHA
†,‡
1.
WEB WEB WEB DoS Denial of Service
J. D. Tygar
††
†,‡‡
CAPTCHA CAPTCHA CAPTCHA CAPTCHA
CMU CAPTCHA
CAPTCHA [1] WEB
A Proposal of CAPTCHA using Strangeness in Machine Translation
Takumi YAMAMOTO†,‡ J. D. Tygar †† Masakatsu NISHIGAKI†,‡‡
CAPTCHA is a technique to prevent automatic programs from being able to acquire free Email or online service accounts. However, as many researchers have reported, the conventional CAPTCHA could be defeated by recent malwares since the ability of PCs get closer to that of human. Therefore CAPTCHA should be based on an even more advanced human cognitive processing ability. In this study, to realize a new CAPTCHA, we propose to use a human ability to recognize “strangeness”. As an example, this paper focuses on strangeness in machine translated sentences, and proposes a CAPTCHA which detects malwares by checking if a user can distinguish natural sentences created by human from machine translated sentences.
† Graduate School of Science and Technology Shizuoka University †† Computer Science Division, University of California, Berkeley ‡ DC1 Research Fellow of the Japan Society for the Promotion of Science (DC1) ‡‡ CREST Japan Science Technology and Agency CREST
CAPTCHA
1 [2] OCR [3]
CAPTCHA CAPTCHA
Figure 1
1 Google CAPTCHA An example of a CAPTCHA used for Google Accounts
Asirra [4]
CAPTCHA Asirra
Asirra [5]
1
2009 Information Processing Society of Japan
ipsjstyle-ms2009.dot (V2.0)
CAPTCHA
PC
CPU Asirra 2
CAPTCHA
CAPTCHA
Figure 2
2 Asirra [4] An example of the authentication window of Assira [4] Asirra
[5] CAPTCHA Asirra Asirra CAPTCHA
2.
CAPTCHA [3] [5] CAPTCHA Asirra [4] CAPTCHA OCR
3.
3.1
2
2009 Information Processing Society of Japan
ipsjstyle-ms2009.dot (V2.0)
NS(Natural Sentence) GS(Garbage Sentence)
3.2.1 NS
NS WEB CAPTCHA
NS NS SS-CAPTCHA NS
WEB
1 in Sentences SS-CAPTCHA SS-CAPTCHA CAPTCHA using Strangeness 3 CS Collected Sentence SS-CAPTCHA NS CS NS SS-CAPTCHA 3.3
3.2.2 GS
GS
SS-CAPTCHA Figure 3
3.2
GS
3 SS-CAPTCHA An overview of our SS-CAPTCHA STEP 1. S0 S0 S0 NS
SS-CAPTCHA
CAPTCHA
3
2009 Information Processing Society of Japan
ipsjstyle-ms2009.dot (V2.0)
S0 STEP 2. L1 STEP 3. L0
L0 (MTL0->L1) L1 STEP 2 S1 (r ) S5 MTL0->L1 (S0) S0 L0 . STEP 3. STEP 2
WEB CS Collected Sentence
STEP 4.
(MT -> (MT (MT -> (MT -> STEP 2 Sr+1 = MTLr-> (Sr) GS (
->
S5 = MT S5 = MT
-> ->
(MT (MT )
-> ->
(MT (MT
-> ->
(S0))))) (S0))))) . GS
STEP 1. CS
NS
GS NS N GS G S=N+G+C NS
Sr CS STEP 2. STEP 3. C
S M N NS
NS
(r)
3.3 SS-CATCHA
SS-CAPTCHA NS 2 1 CAPTCHA NS SS-CAPTCHA
2
STEP 4. STEP 3 CS CS 2 CS CS GS 4 6 GS CS CS K CS K K CS NS CS GS GS NS GS GS V CS GS V 0 CS V 1 V STEP 2 V 1
CS
STEP 4
CS
STEP 1. [a]
V NS
GS
CS
STEP 2.
STEP 1 WEB
WEB
GS
GS
a
4
2009 Information Processing Society of Japan
ipsjstyle-ms2009.dot (V2.0)
4.
SS-CAPTCHA 8
4.1
SS-CAPTCHA 4 SS-CAPTCHA Figure 4 The scheme of SS-CAPTCHA. culture
flickr [6] food
sports
1
Figure 5
5 The scheme of collecting phase SS-CAPTCHA.
8 45.11 GS GS [7]
161
[b ]
1 SS-CAPTCHA 1
3
1 161 Figure 6 6 The scheme of authentication phase in SS-CAPTCHA
b 5
1
GS 161 322
1 GS
2
GS ,
A
20
1
21
2009 Information Processing Society of Japan
ipsjstyle-ms2009.dot (V2.0) NSs NS B A GS STEP 2. NS NS Table 1 1 GS Quality of Garbage Sentence created by processing technique
A 16 B 15 A 23 B 11 A 54 B 14
STEP 1.
5 10
NS
N=5 G=10 GS 1 7 5
GSs(1) NS GS 1 GS 15 15 2
GS GS STEP 1 4.1 A NS 2 GS
~
STEP 3. STEP 4. M=5 STEP1~STEP4 15 1 2
GS 5 GS
10 30
45.11
15 5 4.1 NS
NS
1 B B SS-CAPTCHA GS A
GS
2 GS NS 3.3
A
5 C GS D GS GS 3.3 NS NS [c] D GS
4.2
GS GS 4.1 , , 4.1 GSs(3) 4.1 NS GS , , GS NSs GSs(1),GSs(2), GS
c
5
GS GS
GS 60 GS 5 NS
NS=5
NS
GS
NS
6
2009 Information Processing Society of Japan
ipsjstyle-ms2009.dot (V2.0)
GS 8 NS GS NS
NS 3.3
NS=5
5.
4
5.1
4 SS-CAPTCHA
45 1 2
60
4.2 3.3 GS 1 1 SS-CAPTCHA [9] Figure 7 7 An example of an authentication window of SS-CAPTCHA CAPTCHA CAPTCHA NS 15
Table 2
2 Authentication success rate and time required for sentence selection
GS C 5 4 17.50% 76.25% 93.75% 100.00% D 52.50% 93.75% 100.00% 100.00% C 46.25% 91.25% 100.00% 100.00% D 71.25% 98.75% 100.00% 100.00% C 28.75% 92.50% 100.00% 100.00% D 71.25% 98.75% 100.00% 100.00%
5.2 5.2.1
4.2 4.2
15 5
5 NS
NS=4
1/15C5 5 4 NS
1/3003 NS
5C4
NS
3 2
/
15C4
1/273
69.36 [sec]
66.73 [sec]
60.37 [sec]
7
2009 Information Processing Society of Japan
ipsjstyle-ms2009.dot (V2.0)
5.2.2
6.
CAPTCHA SS-CAPTCHA GS
SS-CAPTCHA
SS-CAPTCHA
NS GS CAPTCHA CAPTCHA CAPTCHA CAPTCHA CAPTCHA [1 10] [9] CAPTCHA CAPTCHA CAPTCHA [9]
GS
5.2.3
No.20-6290
CAPTCHA
SS-CAPTCHA
CAPTCHA CAPTCHA
[1] CAPTCHA
CAPTCHA
1) The Official CAPTCHA Site, http://www.captcha.net 2) PWNtcha-Captcha Decoder, http://caca.zoy.org/wiki/PWNtcha 3) J.Yan,A.S.E.Ahmad: Breaking Visual CAPTCHAs with Naive Pattern Recognition Algorithms, 2007 Computer Security Applications Conference, pp.279-291,2007. 4) J.Elson,J.Douceur,J.Howell,J.Saul:Asirra: a CAPTCHA that exploit interest-aligned manual image categorization. 2007 ACM CSS, pp.366-374, 2007 5) P.Golle:Machine Learning Attacks Against the ASIRRA CAPTCHA, 2008 ACM CSS, pp.535-542 2008. 6) Welcome to Flickr - Photo Sharing, http://www.flickr.com/ 7) , http://www.excite.co.jp/world/ 8) 4 CAPTCHA 2009 CD-ROM No.3D3-3 2009 9) Google CAPTCHA ZDNet Japan http://japan.zdnet.com/sp/feature/07zeroday/story/0,3800083088,20392346,00.htm?ref=rss 10) -CNET Japan http://japan.cnet.com/news/sec/story/0,2000056024,20065869,00.htm
8
2009 Information Processing Society of Japan