HAZE In The Eye of Social Media

2015 edition

Given that we all need air to survive and air is all around us, the haze was a unique event that affected all people in Singapore. At the time, it seemed as if the haze was all we could think about. Local media covered almost nothing but haze for several days. Office chatter and neighborly greetings were abuzz with talk of the haze. Some people took to checking the PSI (Polution Standard Index) readings obsessively.

Our personal observations suggested that haze was on everyone’s mind. But is there a way to measure the consciousness of Singaporeans during the haze to know their thoughts and feelings? Is there a way to quantify the impact of the haze on human activity?

In this data research work, LARC combines the knowledge and expertise of researchers from SMU School of Information Systems and School of Social Science to study social media usage by Singapore users during the haze event so as to derive some insights about people’s reactions to the haze. We analyze several aspects including tweet activities, topics, emotional states, sources of information, and physical behaviors.

Tweet Activity
All Tweets

Loading ...
Proportion of HazeTwitterData

Loading ...

Users
Number of Users generating HazeTwitterData

The figure shows the number of users generating HazeTwitterData.

Loading ...

Topics
Topics on HazeTwitterData

The top frequent words that appear in HazeTwitterData were manually determined and categorized into four topical categories, namely:

  • PSI category: "nea", "psi"
  • n95 mask category: "mask", "n95"
  • forest fire category: "forest", "fire", "hotspots", "hotspot", "indonesia", "sumatra", "riau", "nature", "smoke", "burn"
  • health category: "asthma", "breathing", "health", "unhealthy", "hospital", "clinic", "doctor", "sick", "respiratory"

Loading ...
Topics on all tweets
Loading ...

Emotions
Emotions on HazeTwitterData

We examined the emotion state of users by classifying the tweets into the categories below according to the emotion associated keywords.

  • Negative categories:
    • Anxiety category: “worried”, “worry”, “worries”, “fear”, “afraid”, “frightened”, “scared”, “stress”, “upset”, “nervous”, “anxious”, “alarm”, “tense”, “distress”, “panic”, “die”
    • Anger category: “mad”, “frustrate”, “irritate”, “annoy”, “hate”, “kill”, “piss”, “mean”, “hostile”, “disgust”
    • Swear category: “piss”, “fuck”, “damn”, “shit”, “crap”, “oh no”, “OMG”, “holy __”, “FML”
    • Low Arousal category: “unhappy”, “miserable”, “sad”, “depress”, “hopeless”, “gloomy”, “tired”, “sleepy”, “lethargic”, “fatigue”, “helpless”, “down”, “dejected”
  • Gratitude category: “thank”, “thankful”, “grateful”, “blessed”, “lucky”, “fortunate”, “pleased”
  • Positive category: “happy”, “love”, “glad”, “pleased”, “relax”, “calm”, “relieve”, “phew”, “inspire”, “proud”, “joyful”, “excite”, “admire”, “cheerful”, “delight”, “eager”, “elate”, “enthusiastic”, “interest”, “peaceful”, “pleasant”, “respect”

Loading ...
Proportion of different emotions on HazeTwitterData
Loading ...
Emotions on all tweets
Loading ...

Content Sources

Top Domains from haze-related tweets in September and October 2015

To determine the information sources referenced by Twitter users during the haze event, we examined the highly mentioned and retweeted URL domains among haze related tweets. The URL domains were categorized into three categories below:

  • News media category: e.g., straitstimes.com, todayonline.com, channelnewsasia.com, etc.
  • Government category: e.g., nea.gov.sg, moe.gov.sg, etc.
  • Others category
Tables below show that the top URL domains mentioned and retweeted by users in September and October 2015 are from mainstream news sites (e.g., straitstimes.com, todayonline.com and channelnewsasia.com), government agency (e.g., NEA), and several other popular social media sites (e.g., facebook.com, instagram.com, youtube.com, etc.). This shows that most users still referred to official news channels for information about haze.

Legend:

Domains from news media are highlighted in blue
Domains from government are highlighted in green
Other Domains are not highlighted

Domains mentioned in September and October 2015

147,465 haze-related tweets contain url

nea.gov.sg50,891
channelnewsasia.com25,512
straitstimes.com13,804
todayonline.com11,311
twitter.com7,173
instagram.com6,511
aqicn.org4,983
facebook.com1,871
sgp.si1,585
webreathewhatwebuy.com1,524
asiaone.com1,165
stomp.com.sg999
blissair.com957
youtube.com920
haze.gov.sg918
thestar.com.my842
yahoo.com713
bbc.co.uk671
astroawani.com549
coconuts.co526
nst.com.my418
ask.fm403
mongabay.com382
trendinalia.com369
bloomberg.com368
eco-business.com334
tnp.sg331
latimes.com310
beritasatu.com294
thejakartapost.com283

Domains retweeted in September and October 2015

108,083 haze-related retweets contain url

nea.gov.sg47,596
channelnewsasia.com22,146
straitstimes.com12,132
todayonline.com9,508
twitter.com3,651
aqicn.org1,806
webreathewhatwebuy.com1,163
stomp.com.sg891
bbc.co.uk540
astroawani.com441
instagram.com428
thestar.com.my412
sgp.si395
facebook.com384
youtube.com357
coconuts.co303
yahoo.com303
asiaone.com256
haze.gov.sg251
mongabay.com251

Check-Ins
Foursquare Check-Ins by location category

4SQData were analyzed to study the impact of haze to user activities and businesses. The chart shows foursquare check-in’s trend in September 2015 by location category. We observe four categories of check-in’s:

  • Food: venues contain food, restaurant, and cafe in their category
  • School: venues contain university, school, and college in their category
  • Shop: venues contain mall, shop, and department store in their category
  • Healthcare: venue contain hospital, clinic, doctor, pharmacy, drug store in their category

Loading ...