search

aitoralmeida / twitter_tests

0 stars 0 forks source link

Identify when a meme/history is "breaking" #21

Open aitoralmeida opened 9 years ago

aitoralmeida commented 9 years ago

To do this we have to implement a metric of "burstiness"

aitoralmeida commented 9 years ago

A metric for this is given in "Investigating Rumor Propagation with TwitterTrails".

aitoralmeida commented 9 years ago

What happens when all the intervals are not consecutive? --> see the following commit, current implementation does not assume that intervals must be consecutive.

aitoralmeida commented 9 years ago

The formula in "Investigating Rumor Propagation with TwitterTrails" doesn't make any sense. Find a new method to identify breaking memes

aitoralmeida commented 9 years ago

Trying the method in "The effects of query bursts in web search"

aitoralmeida commented 9 years ago

This new method makes more sense. Still have to define a base level to identify a burst. Discuss it with @porduna