Tool aims to use Twitter history to predict the future

Published: Jul. 27, 2021 at 8:10 PM EDT
Email This Link
Share on Pinterest
Share on LinkedIn

CHARLOTTE, N.C. (WBTV) - Social media could help us predict the future one day. Researchers at the University of Vermont have developed a tool called the StoryWrangler.

At first glance, it just looks like a bunch of graphs.

This data is telling us a story, though.

It harnesses words and phrases used on Twitter and follows their trends over time.

The idea here is that it can paint a picture of history for us by showing us when major events happened. Then, we can use that history to look to the future.

Jamie Boll talked to Peter Sheridan Dodds, one of the developers, about how it works.

Jamie: Is there an easy way to explain how you’re doing it?

Peter: Imagine going into a forest and you count all the trees, and all the types of trees, and count how many there are of each one. And then you do that every year for thousands of years. Then you’d have this sort of picture of how that forest changes over time, and you say, ‘Which trees are the most important ones this year?’ And you deal with this species, and so on. We do the same thing, but the trees are words. So we’re counting up words, right? So political figures, music performers, and so on – we look at that. And of course, it’s a much faster time scale, so we’re doing it at sort of a 15-minute timescale.

Jamie: Explain more - what’s the purpose of this? What do you want to see it being used for?

Peter: Well, there’s a lot of possibilities. Journalism is definitely one of them. Computational history. Just - what happened?  It’s only Twitter, but Twitter really contains kind of an enormous amount of stuff. It has some weird things in it, but big events in the world have been talked about on Twitter massively. So, it helps us look at the record of what happened and when it happened. It can be hard to remember what happened around a particular event or with a particular character. I think if we look at politics over the last four or five years, there’s been a lot of stuff happening. It’s been turbulent, so it at least gives us sort of an anchor, a bedrock that we can all go back on and say, ‘OK. These are the things that happened.’ It’s not about opinion. That’s a different piece. This is just which events happen.

Jamie: So is it a historical tool rather than a predictive tool?

Peter: Well, prediction comes after that, right? So that depends on how clever you are with the data, potentially. Social prediction is very hard, I think it’s fair to say. But, if we can at least describe what has happened, what is happening now, then we have a chance to predict what’s in the future. I think with social phenomena the end game there is we’ll be able to say ‘Look, this is how much we can predict about tomorrow.’ It might be very little, but we’ve gone a long way with the weather. Now we’ll say you know two weeks from now, we have an idea is gonna get colder. But we can’t really tell it’s going to rain in two weeks, but we know that scientifically, we know that’s the end.

Jamie: And this is available for people to go check out for themselves right now, right?

Peter: Yeah, it’s called. So we’re wrangling stories, we’re not making them. Prediction is hard. Like I said, we’re just trying to like kind of muster them a little bit and sort them out.

Jamie: And where do you see it going next?

Peter: Well, we we’re going to put a lot of different things on there. We’ll put have books on there, but all sorts of other kind of quirky things. Even forest data can go on here. It can be really spread out of the different things. Even things like baby names, how they’ve changed over time, so it’s actually a really general tool for any kind of ecology of things that are changing over time, but there are many different types of markets as well.

Copyright 2021 WBTV. All rights reserved.