Discovering finer-grained signals, however, such as those often associated with data about minorities, can be much harder. Luciano Floridi, a philosopher, addresses this point in a recent paper of his. He says,
The real, epistemological problem with big data is small patterns. […] [But] small patterns may be significant only if properly aggregated. So what we need is a better understanding of which data are worth preserving.
In other words, fine-grained patterns may not be readily visible using existing computational techniques. He continues,
And this is a matter of grasping which questions are or will be interesting. […] [T]he game will be won by those who ‘know how to ask and answer questions.’
Again, this underscores the need for social scientists, who are trained to ask and answer important questions about society; however, it also highlights one of my biggest concerns about some of the big data research and development coming out of the computer science community.
via Big Data, Machine Learning, and the Social Sciences | Medium.
You must be logged in to post a comment.