Tweets Show Hardware Stores Are Disgusting, Motels Are Joyful

A study of over 1.5 million tweets a year shows that people in San Francisco are most disgusted when they are in hardware stores, while Londoners are happiest in motels.

Your tweets can show when and where you are angry, disgusted, sad or happy.

Believing that the tweets contain “an abundance of information” about human behavior, Panote Siriaraya of the Kyoto Institute of Technology, Japan, and colleagues analyzed the content of tweets posted in San Francisco and London from early September 2016 to late August 2017.

Siriaraya says they used data from this period because Twitter removed precise location tags from its posts in 2019. The researchers also already analyzed tweets posted between 2016 and 2017 for a previous study, which they reused.

The team drew a boundary around each city, analyzing every tweet posted in those geographic areas that wasn’t sent by apps like Foursquare, which automatically posts tweets when a user “signs up” with its app.

Overall, the researchers analyzed 1.57 million tweets from Londoners posted by 180,000 users and 390,000 tweets from San Francisco posted by 65,000 users. The researchers said these cities were chosen because they can be accurately tracked using OpenStreetMap, a tool that can determine where tweets were sent.

An artificial intelligence known as a neural network classified the content of each tweet according to the most expressed emotion.

The results show that angry tweets were most common on Wednesdays among San Francisco residents, compared with Tuesdays among Londoners. According to the researchers, this could be explained by the fact that people feel frustrated during the work week.

Tweets posted over the weekend tend to display more positive emotions, such as joy, compared to more commonly expressed negative emotions, such as sadness, on weekdays, Siriaray says.

In San Francisco, people tweeted in disgust when they were near real estate agents’ offices and tended to get angry on the city’s bridges. Siriaray said this anger could be explained by the limited number of bridges in San Francisco, which causes traffic to increase as people move in and out of the city.

In London, anger was expressed at bus stops, possibly out of frustration with waiting for public transport, the researchers said.

According to David Ellis of the University of Bath, UK, the study was carried out on a scale never seen before. However, Ellis says he would like the researchers to reach out to some of those who tweeted to confirm that the neural network correctly classified the emotion behind their post.

Barry Smith of the University of London’s School of Advanced Study also has questions about classifying tweets. “One of their emotions is anticipation,” he says. “It’s funny because you can anticipate events with fear, or you can anticipate them with pleasure.” In their study, the researchers considered expectation to be a positive emotion.

Smith also suggests that some places are categorized in such a way that it’s hard to know what’s going on around the person posting. For example, tweets containing the word “wine” were associated with sadness in San Francisco and fear in London.

Siriaraya says that the results and conclusions that can be drawn from them do not always reflect the complexities of human life and require further research.