Best solution for fetching tweets mentioning hundreds of terms?


#1

Hi, I need a view with the latest tweets that includes any term of a list of 400-500 terms. I’ve tried with tools like Hootsuite and Tweetdeck (search with OR operators), but they don’t work for this and I’ve been told that it’s a problem with API limits that I would experience with any external tool.

So I’m thinking of creating a very basic tool myself (only for my personal use) that somehow meets my requirements without going over API limits. I don’t need real time results, an update of the tweets every 3-4 hours could be enough, but I need the results to be as complete as possible (all the tweets mentioning those terms, not a limited sample). The terms I want to track are not very popular, most of them have less than 50 tweets each day.

I’ve read https://dev.twitter.com/rest/public/search and https://dev.twitter.com/streaming/overview but still not sure if the best solution for me is Search API or the Streamings API (Public stream I guess).

I’m not really a developer. If the documentation is good, I can manage to use Github libraries (mostly PHP) and things like that, but I’m looking for the easiest solution possible.

So, what are your recommendations? Search API or Streamings API? Any good tool or library for someone with very limited programming skills? (I’ve seen several possible solutions in Github, including Twitter’s official Hosebird client, but I don’t have any idea on how to choose one over another).

Thanks a lot for your help!


#2

The issue here is the number of terms in your list, which is at the edge of what the Streaming API can support:

The default access level allows up to 400 track keywords, 5,000 follow userids and 25 0.1-360 degree location boxes.

The Search API is intended for short-term real-time access to a few days of data; the Streaming API is intended for ongoing tracking of a range of terms, usernames or locations, but up to 1% of the total tweet volume (so in fact you’re probably fine with your <50 terms per day).

I’d give the Streaming API a go with the filter endpoint passing in a track parameter, but it will top out at maximum 400 terms.


#3

hi,
do you have suggestions of sorting out tweets which have been streamed for several tracking keys, so that tweets for a particular tracking key can be isolated from another?