Towards a new paradigm for segregation measurement in an age of big data