No. 161 Sentiment Extraction from Live Tweets 2014 3
Twitter Twitter Summary Recently, microblogs such as Twitter become popular, and we can tweet about our own daily life easily. The user who tweets during watching sports programs. dramas, and movies on TV becomes increase. In this way, users can share their sentiment with other people who watch the same program on TV by tweeting some scene of the program in real time. Furthermore, user has different sentiment on the timeline based on a scene of a program. In this paper, we propose extracting users sentiment from live tweets. In this time we use the tweets which tweet about movie programs on the TV. Tweets have emoticon and special words. Users often use them to present their sentiment on the tweet. Then we also extract sentiment of the emoticon and special words and we include these sentiment to the tweet sentiment.
1 1 2 3 3 5 3.1... 5 3.1.1 Web... 6 3.2... 6 3.3... 8 4 8 4.1... 8 4.2... 13 4.3... 13 4.4... 16 4.5... 16 5 19 5.1... 19 5.2... 22 5.3... 22 5.4... 22 5.5... 24 5.6... 24 5.6.1... 26 5.6.2... 26 6 28 6.1... 28 6.2... 29 7 30
1... 1 2... 7 3... 9 4... 10 5... 10 6... 10 7... 14 8 7... 14 9 7... 14 10... 15 11 3... 17 12... 18 13 Juman... 20 14... 25 15... 29 16... 30
1... 4 2... 5 3... 6 4... 11 5... 12 6... 17 7... 19 8... 20 9... 21 10... 23 11... 24 12... 26 13... 27 14... 27
1 Twitter 1 Twitter 140 Twitter 1 2 1,000 Facebook 19.2 mixi 19.5 Twitter 37.2 Twitter 1: 1 Twitter https://twitter.com/ 2 http://bit.ly/10e93qn 1
1 (ˆoˆ) (ˆoˆ) 3 (ˆoˆ) (ˆoˆ) (ˆoˆ) 3 2 3 4 5 6 2
2 Plutchik[1] 8 4 [2] [3] [4] [2] 10 6 6 ( ) [5] ( ) neutral 11 [6] [7] Twitter [8] Twitter Ekman [9] Twitter ( ) 4 [10] [11] 1 1 [20] [21] [22] [24] [28] [29] 3
1: + [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [20] [28] [29] [30] [31] [12] [13] [14] [16] [30] [12] 4 [13] 3 [30] 14 5 [14] + [32] ( ) 3 4
3 3.1 6 [3][33] 2 6 6 3 1 2 2 2 3 3-3 3 2 2: 5
3.1.1 Web Twitter 100 Web Web URL 4 2 3.2 [3] 3 PHP Twitter Search API 3 3: 8124 2616 1666 2742 1084 1. (RT). 2. # URL 6
2: 7
3. 4. (3) 5. (1) (4) 3.3 4 5 6-3.0 3.0-1.0 1.0 1-3.0 3.0 4 Twitter (ˆoˆ) orz Twitter 140 4.1 5 6 4 3 174 5 8
3: 9
4: 5: 6: 10
4: 5 1 270 5 2 11
5: 12
4.2 7 7 8 9 7 (T-T) 0 0.92 (;o;) 0.62 0.25 ( ) 0.4 ( ) 0.5 4.3 (ˆoˆ) 10 (ˆoˆ) (ˆoˆ) (ˆoˆ) [34] 3 ( ) 5 270 13
7: 8: 7 9: 7 14
10: 3 11 11 (*ˆˆ*) ((((; )))))) ( ˆ ˆ ) ( ˆ ˆ ) (* ` *) ( ) ( ` ) (T T) ( ) 4 2 (ˆoˆ)!! ( ) 15
( `) 6 4.4 11 TIM e i = DIMi α (1) TIM s j = DIMj β (2) TIM r k = DIMk γ (3) i j k TIM e i i TIM s j j TIM r k k DIMi i DIMj j DIMk k α, β, γ α>γ>β 12 1 4.5 4.3 3 F 7 α =2.0 β =0.2 γ =0.3 F 33% 35% F F 16
11: 3 6: (ˆOˆ)/ = (* `*) + ( ` ) (; ;) +( ) ( ) + ˆˆ 17
12: 18
7: 33% 46% 40% 38% 11% 25% 32% 47% 20% 43% 34% 53% 53% 42% F 39% 28% 42% 36% 18% 34% 33% 38% 64% 44% 36% 13% 25% 37% 63% 22% 58% 21% 56% 47% 45% F 48% 33% 50% 27% 21% 32% 35% 5 5.1 [3] Twitter 3 [2] Yahoo! 3 74,000 8 10 1 2 Plutchik [1] 1-1.0 1.0-1.0 1.0 9 Juman Juman 13 / 3 Yahoo! http://movies.yahoo.co.jp/ 19
8: 13: Juman 20
9: 0.9266 0.0142 0 0.3717 0.500 0.9091 0.609 0 0.1341 0.500 0.9086 0.6669 0.3142 0.0614 0.3976-0.9498-0.4169-0.4684 0.4789 0.2922-0.9433-0.0428-0.7228-0.3922 0.5964-0.9421-0.179 0.183 0.7351 0-0.9297-0.6949 0-0.1439-0.2434-0.9171 0.2987 0.3142-0.073 0.792 0.1267 0.8927-0.0176-0.1385 0.0561-0.2594 0.8809 0.0507-0.2463 0.2473 0.2212 0.8796 0.2179-0.1072-0.0617-0.1873-0.9868 0.3695-0.2085 0.0979-0.506-0.9648 0-0.6922-0.4651-0.2417-0.9616 0 0.0424-0.0938-0.1591 0.0414 0.9713-0.1843 0.1148 0.3598-0.0033 0.7422 0.2494 0.0339-0.1431 0.112 0.7055-0.2193-0.1221 0.0488-0.0319-0.9926-0.1072 0.3976-0.3879 0.1115-0.9926-0.0089 0.2922 0.1349-0.2027-0.9227-0.4284-0.1041-0.0696-0.8497 0.2735 0.9426-0.3096 0.0284 0.1238-0.1801 0.9403 0.1148-0.3405 0.1049 0.0937 0.9051-0.1221 0.0395-0.9047 0.4141-0.9899 0.6284-0.3423 0.4486 0.1794-0.9663 0.0294-0.2891-0.2304 0.0375-0.9662 0.14-0.0481 0.1551 0.183 0.231 0.9697-0.1472 0.4496 0-0.7872 0.8327-0.5429-0.3513 0 0 0.783-0.0081 0.1663 0.207 0.003-0.9906-0.227 0.2675-0.2057 0.004-0.99-0.527-0.1674 0.2563 0.2164-0.8981 21
5.2 [32] 1. 2. 3. 4. 5. 5.3 3.2 3 50 20 11 1. 50 10 3 10 2. (1) (1) 5.4 TW i,j = 1 h (R NR) (4) h 10 i=1 TW i,j i j h R j i 22
10: ( ) 67% 2% 31% 35% 10% 55% 27% 2% 71% 29% 10% 61% 80% 2% 18% 33% 14% 53% 100% 0% 0% 84% 4% 12% 8% 41% 51% 14% 0% 86% NR j i TW i,j j TW i,j j Twitter 10 TW i,j 10 10 4 11 11 7 2 11 70% 2 88.9% 23
11: 7 3 6 2 72.9% 14.3% 12.9% 45.5% 27.2% 27.3% 23.5% 20.6% 55.9% 27.0% 29.7% 43.2% 49.4% 31.6% 19.0% 20.8% 37.5% 41.2% 71.4% 24.5% 4.1% 20.5% 31.3% 48.2% 4.3% 26.1% 69.6% 0% 11.1% 88.9% 5.5 10 i TW i TW i =(SS i SW i )/ max(ss i,sw i ) (5) (5) SS i i ( ) SW i i ( ) TW i 0.97 0.71 0.98 0.58-0.80 1.00 0.93 0.66 1.00 0.95 14 1. 2. 3. 4. 5.6 2 24
14: 25
5.6.1 2013 12 31 19 30 20 1,032 103 5 12 12: 3 100.0% 17 0% 18 77.8% 2 0% 4 0% 16 0% 6 0% 14 0% 18 94.9% 2 50.0% 3 3 5.6.2 8 54 1. 10 3 10 26
2. 3. F 13 14 13: F 0.000 0.000 0.000 0.400 0.111 0.174 0.417 0.139 0.208 0.308 0.333 0.320 0.286 0.067 0.108 0.000 0.000 0.000 0.333 0.111 0.167 0.222 0.063 0.098 0.261 0.231 0.245 0.133 0.091 0.108 14: F 0.172 0.357 0.233 0.636 0.389 0.483 0.654 0.472 0.548 0.370 0.833 0.513 0.167 0.133 0.148 0.429 0.091 0.150 0.417 0.278 0.333 0.444 0.250 0.320 0.451 0.885 0.597 0.313 0.455 0.370 13 14 F 27
F F F F -0.80 F 6 ONE PIECE FILM Z 5549 6.1 1. 2. 1 3. 1 15 15 28
15: 6.2 1. 2. 1 1 3. 16 5 16 2 2 29
16: 7 3 3 10 30
Yahoo! Twitter 3 Twitter Twitter 24 4 26 3 31
2 24 2012. 5 (DEIM2013), 2013. ARG Web (ARG SIG-WI2) 2013. 158 2013. 6 (DEIM2014), 2014(to appear). 2014 2014(to appear). 32
5 (DEIM2013), 2013. 6 (DEIM2014), 2014(to appear). [1] R.Plutchik The nature of emotions American Scentist Vol.89 pp 344 355 2011 [2] 1993 [3] Vol. J94-D No.3 pp 540 548 2011 [4] Web (WebDB2011) 9 pages 2011 11. [5] Web, Vol.50 pp 1365 1374 2009 [6] (TOD) Vol.46 No.SIG10(TOD28) pp.59 71 2005 [7] 19 (WISS2011) 2011 [8] Twitter 2012 [9] FIT 2011 pp 291 294 2011 [10] Twitter TV 26 IC-R-5-3 2012 [11] Twitter 3 (DEIM Forum 2011) F5-6 2011 33
[12] 37, pp 173 176, 2003 [13],, : 22(2), pp 35 43, 2006 [14],. DD, [ ] 2012-DD-85(1), 7 pages, 2012 [15], (4), pp 135 140, 2004 [16],,, 13, pp 49 55 2006 [17],, 51(3), pp 770 783, 2010 [18],,. 2007(6), pp 13 18, 2007 [19],. DE, 106(150), pp 145 150, 2006 [20],,,. 2003(23), pp 169 176, 2003 [21] M.Ptaszynski - - 53(3), pp 204 210, 2012 [22] CHO HEERYON,,,,. ICS,[ ] 2006(110), pp 1 8, 2006 [23],,,,, II. GN, [ ] 2008(31), pp 97 102, 2008 [24] 7, pp 67 82, 2008 [25],,,, Ho,Tu Bao, Dam,Hieu Chi 18 (NLP2012),pp 1204 1207,2012 34
[26], 2011(1), pp 265 267, 2011 [27] : KARP 6, pp 29 46, 2004 [28],. GN, [ ] 2008(31), pp 127 132, 2008 [29],,, : 22(4), pp 31 39, 2007 [30], 1, pp 151 153, 2010 [31],,. [ ] 2008(69), pp 51 56, 2008 [32] 5 (DEIM Forum 2013) A4-4 2013 [33] No.3 pp 540 548 2011 [34] 17 pp 1155 1158 2012 35