[Air-L] Methods to infer Twitter data location

[Libby Hemphill]

Depending on your tolerance for error and how specific you need the
location to be, you may be able to use the user's location (from their
profile, available through the users API if you still have access) rather
than the tweet's geolocation. People usually tweet from "home".

Libby

On Wed, Feb 22, 2023 at 2:38 PM Shulman, Stu via Air-L <
air-l at listserv.aoir.org> wrote:

> User location on Twitter is very tricky and unreliable. Users must opt-in
> to share geo or to "claim" a location when they set up or edit their
> account profile. Claims of location range from 100% authentic to
> preposterous satirical banter and everything in between. Generally, if you
> restrict yourself to the study geo-located or location-purported data, you
> are looking at a small fraction of the whole dataset. For example, here are
> the top most common 30 "claimed" locations in a dataset of Tweets about
> "Dominion Voting Systems" from November 2020 where the most common from
> 421,000 Tweets is 15,000+ (United States) and the 30th is almost 700 (New
> York, NY). There are 239,051 purported locations in this dataset.
>
> United States
> USA
> Texas, USA
> Florida, USA
> California, USA
> Texas
> Georgia, USA
> Canada
> Venezuela
> New York, USA
> Pennsylvania, USA
> Arizona, USA
> Los Angeles, CA
> Michigan, USA
> Ohio, USA
> North Carolina, USA
> Florida
> New Jersey, USA
> Washington, DC
> Houston, TX
> Virginia, USA
> California
> Tennessee, USA
> South Carolina, USA
> Atlanta, GA
> Washington, USA
> Earth
> Illinois, USA
> Chicago, IL
> New York, NY
>
> To study data like this you may need to concatenate things that are similar
> but the same in practice at the geographical scale you are working with.
> Some locations are just Twitter handles. Others are Gab or Parler handles.
> Some look like this:
>
> A Desk in Open Office Hell
> A Field of White Roses
> A Galaxy Far Away
> A Galaxy far far away
> A Galaxy far, far South
> A Garden of Feelings &Hot Air
> A Getsemani, Jerusalen
>
>
> On Wed, Feb 22, 2023 at 11:32 AM Deena Abul-Fottouh via Air-L <
> air-l at listserv.aoir.org> wrote:
>
> >    Hi aoir,
> >
> >    Could anyone please share resources on methods/applications to infer
> >    Twitter data locations from non-geo-located messages?
> >
> >    Thanks.
> >
> >    Deena
> >
> >    ____________________________
> >
> >    Deena Abul-Fottouh, PhD.
> >
> >    Postdoctoral Research Fellow
> >
> >    Digital Society Lab
> >
> >    Political Science Department
> >
> >    Phone icon t phone: [1](905) 525-9140
> >    Envelope icon email: [2]abulfodm at mcmaster.ca
> >
> >    McMaster University
> >
> >    1280 Main Street West
> >
> >    Hamilton, ON L8S 4M4
> >       McMaster University Brighter World logo
> >
> > References
> >
> >    1. tel:+19055259140
> >    2. mailto:abulfodm at mcmaster.ca
> > _______________________________________________
> > The Air-L at listserv.aoir.org mailing list
> > is provided by the Association of Internet Researchers http://aoir.org
> > Subscribe, change options or unsubscribe at:
> > http://listserv.aoir.org/listinfo.cgi/air-l-aoir.org
> >
> > Join the Association of Internet Researchers:
> > http://www.aoir.org/
> >
>
>
> --
> Dr. Stuart W. Shulman
> Founder and CEO, Texifter
> Editor Emeritus, *Journal of Information Technology & Politics*
> _______________________________________________
> The Air-L at listserv.aoir.org mailing list
> is provided by the Association of Internet Researchers http://aoir.org
> Subscribe, change options or unsubscribe at:
> http://listserv.aoir.org/listinfo.cgi/air-l-aoir.org
>
> Join the Association of Internet Researchers:
> http://www.aoir.org/
>