Tweets reveal hardware stores cause disgust but hostels bring joy

A examine of almost 2 million tweets over one 12 months suggests that folks in San Francisco really feel disgusted when at {hardware} shops and Londoners are most joyful at hostels or motels

Technology

1 February 2023

By Chris Stokel-Walker

Your tweets might reveal when and the place you’re indignant, disgusted, unhappy or joyful.

Believing that tweets maintain a “wealth of information” on human behaviour, Panote Siriaraya on the Kyoto Institute of Technology, Japan, and his colleagues analysed the content material of tweets posted in San Francisco and London from early September 2016 to late August 2017.

Siriaraya says they used knowledge from this era as a result of Twitter eliminated exact location tagging from its posts in 2019. The researchers had additionally already analysed tweets posted between 2016 and 2017 for a earlier examine, which they reused.

The group drew a boundary round every metropolis, analysing each tweet posted inside these geographical areas that wasn’t despatched by apps resembling Foursquare, which tweets robotically when a person “checks in” utilizing its app.

Overall, the researchers analysed 1.57 million tweets from folks in London, posted by 180,000 customers, and 390,000 tweets from San Francisco, posted by 65,000 customers. According to the researchers, these cities have been chosen as a result of they are often precisely traced utilizing OpenStreetMap, a software that may find the place tweets have been despatched.

An synthetic intelligence referred to as a neural community labeled the content material of every tweet based on the emotion it most expressed.

The outcomes counsel that tweets with an indignant tone have been mostly posted on Wednesdays amongst folks in San Francisco, in contrast with Tuesdays for Londoners. This could also be defined by folks feeling annoyed whereas getting by way of a working week, based on the researchers.

Tweets posted on weekends tended to show extra constructive feelings, resembling pleasure, in contrast with extra generally expressed detrimental sentiments, like disappointment, on weekdays, which is of little shock, says Siriaraya.

In San Francisco, folks mostly tweeted posts of disgust when near actual property agent places of work, adopted by {hardware} shops. They additionally tended to be angrier across the metropolis’s bridges. According to Siriaraya, this anger could also be defined by San Francisco’s restricted variety of bridges inflicting visitors to construct up as folks enter and go away town.

In London, anger was expressed at bus stops, maybe reflecting frustrations when ready for public transport, based on the researchers. People in London are most joyful in vacationer lodging venues, resembling motels and hostels, the outcomes counsel. Hotels and eating places have been linked to each anger and pleasure.

The analysis was carried out at a scale that hasn’t been finished earlier than, says David Ellis on the University of Bath, UK. However, Ellis says he would have favored the researchers to method a few of those that tweeted to verify the neural community accurately labeled the emotion behind their put up.

Barry Smith on the School of Advanced Study, University of London, additionally has questions in regards to the classifications of the tweets. “One of the emotions they had is anticipation,” he says. “That’s a funny one, because you can anticipate things with dread or you can anticipate things with pleasure.” The researchers deemed anticipation to be a constructive emotion of their examine.

Smith additionally means that a few of the places are labeled in a approach that makes it troublesome to discern what is going on across the individual posting. For instance, tweets containing the phrase “wine” have been related to disappointment in San Francisco and worry in London.

Siriaraya says that the outcomes, and the insights that may be obtained from them, don’t all the time mirror the complexities of human life and require additional analysis.

More on these matters: