La Nación’s Nisman coverage: Using machine learning to identify significant voicemails

29/2/2016 La Nación’s Nisman coverage: Using machine learning to identify significant voicemails By Juan Elosua and Francis Tseng Back in 2015, Argentinian prosecutor Alberto Nisman was found dead under suspicious circumstances, just as he was about to bring a complaint accusing the Argentinian President Fernández of interfering with investigations into the AMIA bombing that took place in 1994 (this Guardian piece provides some good background). La Nación had access to some 40,000 phone calls related to the case, and wanted to explore them further—but naturally, that is quite a big number, and it’s hard to gather the resources to comb through that many hours of audio. La Nación crowdsourced the labeling of about 20,000 of these calls into those that were interesting and those that were not (e.g. voicemails or…


Link to Full Article: La Nación’s Nisman coverage: Using machine learning to identify significant voicemails