3

I am a student of economics and I am conducting my master thesis on cultural industries. Trying to understand what is the feeling of the people about different themes, I'am using TwitteR package in R and in particular the function searchtwitter. everything works fine but when I am trying to geocode the tweets(geocode is an argument of the function) the results provided are very few. I have tried with different coordinates and setting also the argument "since" and "until" but significant results are obtained only if I don't set the argument "geocode", "since", "until". Poorly speaking I obtain a relevant number of tweets but I don't know their provenience. Do you know some strategy to deal with this issue? Are there better programs to map feelings posted by people on social network?

  • I know it's been done before ... there was ESIP's 'Air Twitter' project looking at air quality : forbes.com/2010/11/05/… . I'd think that people posting pictures would be useful as so many cell phones leak location metadata. You might also look into how Auorasaurus determines location : aurorasaurus.org – Joe Sep 26 '15 at 23:52
3

There are 3 things that are decreasing your search results count:

  1. The Twitter Search API only gives results about 1 week back.

  2. Only a fraction of tweets are geo-tagged.

  3. Search terms may be too specific.

There isn't too much to do about 1 and 2, but for 3, I can recommend getting familiar with Advanced Search. You can construct a query with the website, then use the same query string in the code. As an example, one of my queries has a query string with 366 characters, mostly OR statements and then a few exclude (-) statements.

If you post some specific examples, maybe the community can help more.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.