ground truth labels for TwitterworldCup2014 dataset about midas HOT 3 CLOSED

For Twitter datasets, we only have a few ground truth events so we cannot directly assign a 0/1 label to all the tweets. However, when anomaly detection algorithms are run on such datasets, we can observe that peaks (tweets aggregated per hour/day/week) correspond well to ground truth events.
If you still need to generate labels, you can try and label tweets containing goal/penalty/injury kind of events as anomalous (label = 1)

from midas.

Thanks!

" label tweets containing goal/penalty/injury kind of events as anomalous (label = 1)"

does it mean that whichever tweets that contain events should be considered as anomalous ,and those tweets which do not contain events as nonanomalous?

do you suggest any other twitter or Facebook dataset for edge stream anomaly detection?

from midas.

Yes, I meant that. Sorry, I am not aware of any other datasets from Twitter/Facebook.

from midas.