ry -Disgusting -Joyful--s NYWWWMAME H B 4 D XW 010203040506070809010011120130140150160170180190200210220230240250260270280290300310320330340350360370 Day Figure 4: Abnormal event detection of 2011. We also provide an interactive version on the following link: http://gana.nlsde.buaaedu.cn/hourly-happy/timeline_2011.html. sad. Moodlens is now online for temporal and spa- FAngry-Disgusting-Joyful--Sad tial sentiment pattern ry, abnormal events detection and illustration and al-time monitoring of senti- ment fluctuations 5. ACKNOWLEDGEMENTS This works was supported by the fund of the State Ke Laboratory of Software Development Environment(SKLSDE 2011ZX-02)and the Research Fund for the Doctoral Pro- gram of Higher Education of China(Grant No. 20111102110019 The third author was supported in part by the National Natural Science Foundation of China(NSFC) under Grants 71171007,70901002,and90924020 6. REFERENCES Figure a sample from Imon- Sho aoki and osamu Uchida. A Method for g online appl automatically generating the emotional vectors of gana. nlsde. buaa. edu. cn/hourly-happy/line- 36h.htmL. emoticons using weblog articles. ACACOS'1l, pages 132-136, Stevens Point, Wisconsin, USA, 201 2 Johan Bollen, Huina Mao, and Xiaojun Zeng. Twitter PM of Dec. 24. 2011 and ends to 7: 00 AM of Dec. 26. 2011 Mood predicts the stock market. Journal of the delay between two points is 30 minutes and 72 point omputational Science, 2(1): 1-8, 2011 re drew in all. As can be seen, the real-time event could 3 Johan Bollen, Alberto Pepe, and Huina Mao Modeling be indeed monitored. For instance, regarding to the time public mood and emotion: Twitter sentiment and marked by A. which is at 0: 00 PM of Dec. 24. the sentiment ocio-economic phenomena. 5th ICWSM, 2011 of joyful reaches peak because everyone is celebrating the 4 Rong-En Fan, Kai-Wei Chang, Cho-Jui Hsieh coming of the Christmas Day. While with respect to B, at ang-Rui Wang, and Chih-Jen Lin. Liblinear: A the time of 1: 00 AM of Dec. 26. the fraction of sad tweets library for large linear classification. J. Mach. Learn rises suddenly, the bi-gram terms reveal that it is caused by Res.9:1871-1874,June2008. the earthquake happened in Chengdu at that time. 5 Scott A. Golder and Michael W. Macy. Diurnal and seasonal mood vary with work, sleep, and daylength 4. CONCLUSION across diverse cultures. Science, 333(6051): 1878-1881 MoodLens is an online sentiment analysis system r Chi- September 2011 ese tweets in Weibo. It employs the emoticons for the gen-[6 ation of sentiment labels for tweets, and builds an incre- http://gana.nlsde.buaa.edu.cn/hourly-happy/line-cycl mental learning Naive Bayes classifier for the categorization 7] of four types of sentiments: angry, disgusting, joyful, and http://gana.nlsde.buaaedu.cn/hourly-happy/map-pie.html0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 Day fraction Angry Disgusting Joyful Sad E F C A D H G I J B Figure 4: Abnormal event detection of 2011. We also provide an interactive version on the following link: http://gana.nlsde.buaa.edu.cn/hourly happy/timeline 2011.html. 0 10 20 30 40 50 60 70 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 time fraction Angry Disgusting Joyful Sad B A Figure 5: A sample from real-time mon￾itoring. The online application could be accessed through the following link: gana.nlsde.buaa.edu.cn/hourly happy/line 36h.html. PM of Dec. 24, 2011 and ends to 7:00 AM of Dec. 26, 2011, the delay between two points is 30 minutes and 72 points are drew in all. As can be seen, the real-time event could be indeed monitored. For instance, regarding to the time marked by A, which is at 0:00 PM of Dec. 24, the sentiment of joyful reaches peak because everyone is celebrating the coming of the Christmas Day. While with respect to B, at the time of 1:00 AM of Dec. 26, the fraction of sad tweets rises suddenly, the bi-gram terms reveal that it is caused by the earthquake happened in Chengdu at that time. 4. CONCLUSION MoodLens is an online sentiment analysis system for Chi￾nese tweets in Weibo. It employs the emoticons for the gen￾eration of sentiment labels for tweets, and builds an incre￾mental learning Na¨ıve Bayes classifier for the categorization of four types of sentiments: angry, disgusting, joyful, and sad. MoodLens is now available online for temporal and spa￾tial sentiment pattern discovery, abnormal events detection and illustration, and online real-time monitoring of senti￾ment fluctuations. 5. ACKNOWLEDGEMENTS This works was supported by the fund of the State Key Laboratory of Software Development Environment (SKLSDE- 2011ZX-02) and the Research Fund for the Doctoral Pro￾gram of Higher Education of China (Grant No. 20111102110019). The third author was supported in part by the National Natural Science Foundation of China (NSFC) under Grants 71171007, 70901002, and 90924020. 6. REFERENCES [1] Sho Aoki and Osamu Uchida. A method for automatically generating the emotional vectors of emoticons using weblog articles. ACACOS’11, pages 132–136, Stevens Point, Wisconsin, USA, 2011. [2] Johan Bollen, Huina Mao, and Xiaojun Zeng. Twitter mood predicts the stock market. Journal of Computational Science, 2(1):1–8, 2011. [3] Johan Bollen, Alberto Pepe, and Huina Mao. Modeling public mood and emotion: Twitter sentiment and socio-economic phenomena. 5th ICWSM, 2011. [4] Rong-En Fan, Kai-Wei Chang, Cho-Jui Hsieh, Xiang-Rui Wang, and Chih-Jen Lin. Liblinear: A library for large linear classification. J. Mach. Learn. Res., 9:1871–1874, June 2008. [5] Scott A. Golder and Michael W. Macy. Diurnal and seasonal mood vary with work, sleep, and daylength across diverse cultures. Science, 333(6051):1878–1881, September 2011. [6] http://gana.nlsde.buaa.edu.cn/hourly happy/line cycle.html. [7] http://gana.nlsde.buaa.edu.cn/hourly happy/map pie.html