This program classifies tweets in Python if it is safe to ignore, possibly concerning, and strongly concerning using Support Vector Machine.
Note: the data is not well refined.
References: Detecting suicidality on Twitter by O'Dea et al, 2015
http://blog.christianperone.com/2011/09/machine-learning-text-feature-extraction-tf-idf-part-i/