diversity.py

# Different functions for diversity may be defined here.
# A diversity function gets a list of objects and return one float value between zero and one.
# A value close to one means higher diversity.
# Input for user groups: redundant list of all members of all groups.
# Input for spatiotemporal data: list of distance values.

# diversity definition for spatiotemporal data
def diversity_calculator_geo(elements):
	count = len(elements)
	sum = 0
	for value in elements:
		sum += value
	average = sum / count
	return average

# diversity definition for user data as jaccard (as in CIKM'15 paper)
def diversity_calculator_jaccard(users_list):
	users_set = set(users_list)
	unqiue_count = len(users_set)
	all_count = len(users_list)
	jaccard = float(unqiue_count) / float(all_count)
	return jaccard

# diversity definition for user data as exponential (as in PKDD'16 paper)
def diversity_calculator_expo(users_list):
	users_set = set(users_list)
	unqiue_count = len(users_set)
	all_count = len(users_list)
	nb_overlaps = all_count - unqiue_count
	diversity_expo = 1.0 / float(1 + nb_overlaps)
	return diversity_expo

def diversity(elements):
	return diversity_calculator_expo(elements)
	# return diversity_calculator_jaccard(elements)
	# return diversity_calculator_expo(elements)