#1

Hey, I ran a filtered stream for one hour using the rule “A or B” for about an hour (A and B being two keywords of interest) and I ran a recent search for the same time period using the same rule “A or B”. I see there’s a difference in the number of tweets extracted from both endpoints. I have checked for the usual suspects like deleted tweets, matching the exact time period for these tweets but the difference is still there. I missed 157 tweets in recent search and 13 tweets in streaming. Just wanted to know whether the completeness for these v2 endpoints is supported or not or am I missing something obvious here?

2 Likes
#2

Thanks for reaching out.
I have a few follow-up questions here:

  • Do you have code samples for both recent search and filtered stream?

  • Do you have examples of what you expected to see but didn’t?

Thanks again!
Jessica

#3

Hey @jessicagarson
Sorry for the late reply…

Code samples

Recent Search function

recent_search
recent_search1305×573 80.6 KB

Filtered Stream function

stream
stream1378×360 48.9 KB

I ran the filtered stream and recent search for the same query for the same time period (first the filtered stream and then the recent search for the stream start and end timestamps as start_time and end_time for recent search). I missed few tweets in both, few were present in stream results while not in recent search results and vice versa. I have mentioned the numbers in my initial post.
Example: 1341741060595339264, 1341742399140327429, 1341746314766413824, 1341747097377226753 are some tweets captured while streaming but not in recent search while 1341742882319978498, 1341745981453381632 were present in recent search and not in stream. I expect both result sets to be the same (adjusting the deleted tweets) for the same time period.

So I wanted to know whether these APIs provide completeness or not, or am I doing something wrong or missing something?