def _get_edges_per_slot():
slots = {} # {year : edges}
for year in range(1992, 2004):
slots[year] = HepReader.read_edges("datasets/hepth/timeslots/cit-HepTh-{0}.edges".format(year))
return slots
P.show()
def plot_data_distribution():
"""Plot number of new papers and new citations per year, take Hep files as input"""
minyear = 1992
maxyear = 2003
x = [datetime(year, 01, 01) for year in range(minyear, maxyear + 1)]
x.extend([datetime(year, 06, 01) for year in range(minyear, maxyear + 1)])
x.extend([datetime(year, 9, 1) for year in range(minyear, maxyear + 1)])
x.extend([datetime(year, 12, 1) for year in range(minyear, maxyear + 1)])
x = sorted(x)
dates = HepReader.read_dates("/home/stpk/dev/role-mining/datasets/hepth/cit-HepTh-dates-cleaned.txt")
edges = HepReader.read_edges("/home/stpk/dev/role-mining/datasets/hepth/cit-HepTh.txt")
slots = HepReader.split_to_timeslots(dates, edges, x)
for slot in slots.values():
print len(slot)
citats = [len(slots[t]) for t in x]
print citats
print x
fig, ax = P.subplots()
ax.xaxis_date()
P.title("Nowe publikacje na kwartal")
P.plot(x, citats, color='b', alpha=0.5, label="Nowe cytowania")
ax.set_xticks(x)
P.xticks(rotation=70)
