gnip now offering geolocation of twitter profiles and text
Hi everyone, for those of you trying to map tweets in your work, I just saw that GNIP is now offering geographic enrichment of their Twitter stream that geocodes both the Location information in user profiles AND the tweet text itself (it separates the two) and makes this available as lat/long data for mapping/filtering/etc alongside the traditional sensor-based and software-set Exact Location and Place geographic fields. http://blog.gnip.com/twitter-geo-data-enrichment/ While GNIP is using their own engine for this, you can see an overview of what the profile and tweet text geography of Twitter looks like in my paper from earlier this year: http://firstmonday.org/ojs/index.php/fm/article/view/4366/3654 Thought I would share this with the list since using GNIP's stream you can now essentially leave the geographic coding to them and just get an additional set of metadata for each tweet capturing what geographic information it contains, making it trivial to map tweets both by where they are coming from and where they are talking about... Kalev Leetaru Yahoo! Fellow in Residence Institute for the Study of Diplomacy Edmund A. Walsh School of Foreign Service Georgetown University
Very interesting. Wonder how they're dealing with the limitations. http://dfreelon.org/2013/05/12/twitter-geolocation-and-its-limitations/ http://takhteyev.org/papers/Takhteyev-Wellman-Gruzd-2010.pdf Alexander Leavitt PhD Student USC Annenberg School for Communication & Journalism http://alexleavitt.com Twitter: @alexleavitt On Fri, Aug 23, 2013 at 1:59 PM, kalev leetaru <kalev.leetaru5@gmail.com>wrote:
Hi everyone, for those of you trying to map tweets in your work, I just saw that GNIP is now offering geographic enrichment of their Twitter stream that geocodes both the Location information in user profiles AND the tweet text itself (it separates the two) and makes this available as lat/long data for mapping/filtering/etc alongside the traditional sensor-based and software-set Exact Location and Place geographic fields.
http://blog.gnip.com/twitter-geo-data-enrichment/
While GNIP is using their own engine for this, you can see an overview of what the profile and tweet text geography of Twitter looks like in my paper from earlier this year:
http://firstmonday.org/ojs/index.php/fm/article/view/4366/3654
Thought I would share this with the list since using GNIP's stream you can now essentially leave the geographic coding to them and just get an additional set of metadata for each tweet capturing what geographic information it contains, making it trivial to map tweets both by where they are coming from and where they are talking about...
Kalev Leetaru Yahoo! Fellow in Residence Institute for the Study of Diplomacy Edmund A. Walsh School of Foreign Service Georgetown University _______________________________________________ The Air-L@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/
Alex, while it is of course impossible to know the true representativeness of Twitter and in particular even the representativeness of users with the different classes of geographic fields, two key findings are that sensor-based tweets (ie, Exact Location tweets via GPS and cellular triangulation) are highly spatially correlated with the distribution of power via the NASA Night Lights imagery, suggesting that there is not a strong macro-level spatial skew in the availability of sensor-based tweets (ie, that Western users are more likely to have GPS tagging of their tweets enabled) and secondly, that the results of geocoding user profiles is, in turn, highly spatially correlated with sensor-based tweets. ~Kalev On Fri, Aug 23, 2013 at 6:31 PM, Alex Leavitt <alexleavitt@gmail.com> wrote:
Very interesting. Wonder how they're dealing with the limitations.
http://dfreelon.org/2013/05/12/twitter-geolocation-and-its-limitations/ http://takhteyev.org/papers/Takhteyev-Wellman-Gruzd-2010.pdf
Alexander Leavitt PhD Student USC Annenberg School for Communication & Journalism http://alexleavitt.com Twitter: @alexleavitt
On Fri, Aug 23, 2013 at 1:59 PM, kalev leetaru <kalev.leetaru5@gmail.com>wrote:
Hi everyone, for those of you trying to map tweets in your work, I just saw that GNIP is now offering geographic enrichment of their Twitter stream that geocodes both the Location information in user profiles AND the tweet text itself (it separates the two) and makes this available as lat/long data for mapping/filtering/etc alongside the traditional sensor-based and software-set Exact Location and Place geographic fields.
http://blog.gnip.com/twitter-geo-data-enrichment/
While GNIP is using their own engine for this, you can see an overview of what the profile and tweet text geography of Twitter looks like in my paper from earlier this year:
http://firstmonday.org/ojs/index.php/fm/article/view/4366/3654
Thought I would share this with the list since using GNIP's stream you can now essentially leave the geographic coding to them and just get an additional set of metadata for each tweet capturing what geographic information it contains, making it trivial to map tweets both by where they are coming from and where they are talking about...
Kalev Leetaru Yahoo! Fellow in Residence Institute for the Study of Diplomacy Edmund A. Walsh School of Foreign Service Georgetown University _______________________________________________ The Air-L@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/
Thanks for the shoutout, Alex. One of my main points was that while the advances in Kalev et al's paper are quite impressive--they provide an order of magnitude more location coverage than GPS tagging alone--representativeness is still a huge issue. There's an important paper yet to be written about systematic differences between the sets of Twitter users whose locations can and cannot be reliably identified--I'd be shocked if the two groups were statistically indistinguishable. ~DEEN On 8/23/2013 7:17 PM, kalev leetaru wrote:
Alex, while it is of course impossible to know the true representativeness of Twitter and in particular even the representativeness of users with the different classes of geographic fields, two key findings are that sensor-based tweets (ie, Exact Location tweets via GPS and cellular triangulation) are highly spatially correlated with the distribution of power via the NASA Night Lights imagery, suggesting that there is not a strong macro-level spatial skew in the availability of sensor-based tweets (ie, that Western users are more likely to have GPS tagging of their tweets enabled) and secondly, that the results of geocoding user profiles is, in turn, highly spatially correlated with sensor-based tweets.
~Kalev
On Fri, Aug 23, 2013 at 6:31 PM, Alex Leavitt <alexleavitt@gmail.com> wrote:
Very interesting. Wonder how they're dealing with the limitations.
http://dfreelon.org/2013/05/12/twitter-geolocation-and-its-limitations/ http://takhteyev.org/papers/Takhteyev-Wellman-Gruzd-2010.pdf
Alexander Leavitt PhD Student USC Annenberg School for Communication & Journalism http://alexleavitt.com Twitter: @alexleavitt
On Fri, Aug 23, 2013 at 1:59 PM, kalev leetaru <kalev.leetaru5@gmail.com>wrote:
Hi everyone, for those of you trying to map tweets in your work, I just saw that GNIP is now offering geographic enrichment of their Twitter stream that geocodes both the Location information in user profiles AND the tweet text itself (it separates the two) and makes this available as lat/long data for mapping/filtering/etc alongside the traditional sensor-based and software-set Exact Location and Place geographic fields.
http://blog.gnip.com/twitter-geo-data-enrichment/
While GNIP is using their own engine for this, you can see an overview of what the profile and tweet text geography of Twitter looks like in my paper from earlier this year:
http://firstmonday.org/ojs/index.php/fm/article/view/4366/3654
Thought I would share this with the list since using GNIP's stream you can now essentially leave the geographic coding to them and just get an additional set of metadata for each tweet capturing what geographic information it contains, making it trivial to map tweets both by where they are coming from and where they are talking about...
Kalev Leetaru Yahoo! Fellow in Residence Institute for the Study of Diplomacy Edmund A. Walsh School of Foreign Service Georgetown University _______________________________________________ The Air-L@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/
_______________________________________________ The Air-L@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/
-- Deen Freelon, Ph.D. Assistant Professor American University School of Communication Office: Asbury 228A dfreelon@gmail.com http://dfreelon.org
participants (3)
-
Alex Leavitt -
Deen Freelon -
kalev leetaru