def draw_topK(search_type=0, bar_type=0):
# testSampleResultFile.t=0.ns=500.r=1.k=10.nw=1
alg_type = ['SPBase', 'SPBest']
ns = 200
radius = 2
ks = [1, 3, 5, 8, 10, 15, 20]
nw = 5
if bar_type==0:
ys = [i*1000 for i in range(13)]
ys = None
y_label = 'Runtime (ms)'
elif bar_type==1:
ys = [i*100 for i in range(5)]
y_label = '# of TQSP Computations'
else:
ys = [i*10 for i in range(6)]
y_label = '# of R-tree nodes accessed'
base_dir = 'D:\\nowMask\KnowledgeBase\\sample_result\\yago2s_single_date\\k_nw\\'
if bar_type==0:
timeBar = Bar(xLabel=r'top-$k$', yLabel=y_label, is_stack=True, title=r'top-k type=' + str(search_type), ys=ys)
for i in range(len(alg_type)):
datas = []
for k in ks:
data = Data.getData(fp=PathUtility.sample_res_path(base_dir, sp=alg_type[i], nwlen=1000000, mds=1000, t=search_type, ns=200, r=2, k=k, nw=5, wf=500, dr=7))
print(data)
datas.append(data)
timeOthers = []
timeSemantics = []
for data in datas:
timeSemantics.append(data.timeSemantic)
timeOthers.append(data.timeOther)
timeBar.draw_bar(i, timeSemantics, hatch=Bar.hatchx, label=alg_type[i])
timeBar.draw_bar(i, timeOthers, ys=timeSemantics, faceColor='white')
timeBar.show()
else:
if 1==bar_type: y_type = 'TQSP'
elif 2==bar_type: y_type = 'RTree'
bar = Bar(xLabel=r'top-$k$', yLabel=y_label, title=r'top-k type=' + str(search_type), ys=ys, y_type=y_type)
for i in range(len(alg_type)):
datas = []
for k in ks:
data = Data.getData(fp=PathUtility.sample_res_path(base_dir, sp=alg_type[i], nwlen=1000000, mds=1000, t=search_type, ns=200, r=2, k=k, nw=5, wf=500, dr=7))
print(data)
datas.append(data)
list1 = []
for data in datas:
if 1==bar_type:
list1.append(data.numTQSP)
elif 2==bar_type:
list1.append(data.numAccessedRTreeNode)
bar.draw_bar(i, list1, label=alg_type[i])
bar.show()
def draw_date_range(label=r'$|q.\delta|$', base_y=None, x_rotation=0, fpath='test.pdf'):
if base_y!=None: ys = [base_y+i*100 for i in range(0, 17)]
# x_txts = [r'$5\times10^8$', r'$1\times10^8$', r'$5\times10^7$', r'$1\times10^7$',
# r'$5\times10^6$', r'$1\times10^6$', r'$5\times10^5$', r'$1\times10^5$',
# r'$1\times10^4$', '0']
# x_txts = [0,3,7,15,30,50,100,150,300,600]
# drs = [0,3,7,15,30,50,100,150,300,600]
drs = [0,3,7,15,30,50,100,150]
x_txts = [0,6,14,30,60,100,200,300]
xs=[i for i in range(len(drs))]
if base_y!=None: chart = LineChart(xs, x_txts, ys=ys, yscale='liner', xLabel=label, title=label, yLabel='Runtime (ms)', xlabel_rotation=x_rotation, fpath = fpath)
else: chart = LineChart(xs, x_txts, xLabel=label, title=label, yLabel='Runtime (ms)', xlabel_rotation=x_rotation, fpath=fpath)
base_dir = 'D:\\nowMask\KnowledgeBase\\sample_result\\yago2s_single_date\\dr\\'
alg_types=['SPBase', 'SPBest']
for alg_type in alg_types:
runtimes = []
for dr in drs:
data = Data.getData(fp=PathUtility.sample_res_path(base_dir, sp=alg_type, nwlen=1000000, mds=1000, t=1, ns=200, r=2, k=5, nw=5, wf=500, dr=dr))
# print(data)
runtimes.append(data.timeTotal)
print(runtimes)
if alg_type=='SPBase':
chart.draw_line(runtimes, r'$SPTR$')
elif alg_type=='SPBest':
chart.draw_line(runtimes, r'$SPTR^*$')
chart.show()
def draw_differ_size(line_type=0, title='Graph Vertex Size (in million)', fpath='test.pdf'):
# testSampleResultFile.t=0.ns=500.r=1.k=10.nw=1
xs = [1, 2, 3, 4]
x_txts = [2.0, 4.0, 6.0, 8.0]
if line_type==0: yLabel='Runtime (ms)'
elif line_type==1: yLabel = '# of R-tree nodes accessed'
chart = LineChart(xs, x_txts, xLabel=r'Graph Vertex Size (in million)', yLabel=yLabel, title=title, fpath=fpath)
alg_types = ['SPBase', 'SPBest']
ts = [0, 1]
sizes = ['2000000', '4000000', '6000000', 'org']
alg_names = [r'$SPTD$', r'$SPTR$', r'$SPTD^*$', r'$SPTR^*$']
base_dir = 'D:\\nowMask\KnowledgeBase\\sample_result\\yago2s_single_date\\diff_size\\'
alg_names_index = 0
for alg_type in alg_types:
for t in ts:
runtimes = []
for i in range(len(sizes)):
data = Data.getData(fp=PathUtility.sample_res_path(base_dir + sizes[i] + '\\', sp=alg_type, nwlen=1000000, mds=1000, t=t, ns=200, r=2, k=5, nw=5, wf=500, dr=7))
print(data)
if line_type==0: runtimes.append(data.timeTotal)
elif line_type==1: runtimes.append(data.numAccessedRTreeNode)
chart.draw_line(runtimes, label=alg_names[alg_names_index])
alg_names_index = alg_names_index + 1
chart.show()
def draw_nw(base_y=None, fpath='test.pdf'):
ys = None
if base_y != None:
ys = [base_y+i*100 for i in range(0, 11)]
x_txts = [1.0, 3.0, 5.0, 8.0, 10.0]
xs=[i for i in range(len(x_txts))]
yLabel = 'Runtime (ms)'
chart = LineChart(xs, x_txts, ys=ys, xLabel=r'|$q.\psi$|', yLabel=yLabel, title=r'|$q.\psi$|', fpath=fpath)
alg_types = ['SPBase', 'SPBest']
search_types = [0, 1]
type = None
nws = [1, 3, 5, 8, 10]
base_dir = 'D:\\nowMask\KnowledgeBase\\sample_result\\yago2s_single_date\\k_nw\\'
for search_index in range(len(search_types)):
for alg_index in range(len(alg_types)):
if alg_index==0:
if search_index==0: type=r'$SPTD$'
else: type=r'$SPTR$'
else:
if search_index==0: type=r'$SPTD^*$'
else: type=r'$SPTR^*$'
runtimes = []
for nw in nws:
data = Data.getData(fp=PathUtility.sample_res_path(base_dir, sp=alg_types[alg_index], nwlen=1000000, mds=1000, t=search_types[search_index], ns=200, r=2, k=5, nw=nw, wf=500, dr=7))
print(data)
runtimes.append(data.timeTotal)
chart.draw_line(runtimes, type)
chart.show()
def draw_n_words(search_type=0):
# testSampleResultFile.t=0.ns=500.r=1.k=10.nw=1
xs = [0.5, 1, 2, 3, 4, 5, 5.5]
x_txts = [1.0, 3.0, 5.0, 8.0, 10.0]
ys = [i*100 for i in range(15)]
alg_type = ['SPBase', 'SPBest']
ns = 200
radius = 2
k = 5
nws = [1, 3, 5, 8, 10]
base_dir = 'D:\\nowMask\KnowledgeBase\\sample_result\\yago2s_single_date\\k_nw\\'
timeBar = Bar(xLabel=r'|$q.\psi$|', yLabel='Runtime (ms)', is_stack=True, title='n_wrods type=' + str(search_type), ys=ys, xs=xs, x_txts=x_txts, y_type='NW')
for i in range(len(alg_type)):
datas = []
for nw in nws:
data = Data.getData(fp=PathUtility.sample_res_path(base_dir, sp=alg_type[i], nwlen=1000000, mds=1000, t=search_type, ns=200, r=2, k=5, nw=nw, wf=500, dr=7))
print(data)
datas.append(data)
timeOthers = []
timeSemantics = []
for data in datas:
timeSemantics.append(data.timeSemantic)
timeOthers.append(data.timeOther)
timeBar.draw_bar(i, timeSemantics, hatch=Bar.hatchx, label=alg_type[i])
timeBar.draw_bar(i, timeOthers, ys=timeSemantics, faceColor='white')
timeBar.show()
def draw_radius(type=0, base_y=600, title='TEST', fpath='test.pdf'):
ys = [base_y+i*50 for i in range(0, 11)]
xs=[i for i in range(0, 4)]
x_txts = [1.0, 2.0, 3.0, 5.0]
chart = LineChart(xs, x_txts, ys=ys, yscale='liner', xLabel=r'$\alpha$-radius', yLabel='Runtime (ms)', title=title, fpath=fpath)
# chart = LineChart(xs, x_txts, xLabel=r'$\alpha$-radius', yLabel='Runtime (ms)')
radius = [1, 2, 3, 5]
ks = [1, 3, 5, 8, 10, 15, 20]
base_dir = 'D:\\nowMask\KnowledgeBase\\sample_result\\yago2s_single_date\\radius_k\\'
for k in ks:
runtimes = []
for r in radius:
data = Data.getData(fp=PathUtility.sample_res_path(base_dir, nwlen=1000000, mds=1000, t=type, ns=200, r=r, k=k, nw=5, wf=500, dr=7))
print(data)
runtimes.append(data.timeTotal)
chart.draw_line(runtimes, 'k=' + str(k))
chart.show()
def draw_k(show_type, base_y=None, fpath='test.pdf'):
ys = None
ys = [i*100 for i in range(17)]
if base_y != None:
ys = [base_y+i*100 for i in range(0, 17)]
x_txts = [1.0, 3.0, 5.0, 8.0, 10.0, 15.0, 20.0]
xs=[i for i in range(len(x_txts))]
yLabel = None
yscale = 'liner'
if 0==show_type:
yLabel = 'Runtime (ms)'
elif 1==show_type:
yLabel = '# of TQTSP Computations'
yscale='log'
elif 2==show_type: yLabel = '# of R-tree nodes accessed'
chart = LineChart(xs, x_txts, ys=ys, yscale=yscale, xLabel=r'top-$k$', yLabel=yLabel, title=r'top-$k$', fpath=fpath)
alg_types = ['SPBase', 'SPBest']
search_types = [0, 1]
type = None
ks = [1, 3, 5, 8, 10, 15, 20]
base_dir = 'D:\\nowMask\KnowledgeBase\\sample_result\\yago2s_single_date\\k_nw\\'
for search_index in range(len(search_types)):
for alg_index in range(len(alg_types)):
if alg_index==0:
if search_index==0: type=r'$SPTD$'
else: type=r'$SPTR$'
else:
if search_index==0: type=r'$SPTD^*$'
else: type=r'$SPTR^*$'
runtimes = []
for k in ks:
data = Data.getData(fp=PathUtility.sample_res_path(base_dir, sp=alg_types[alg_index], nwlen=1000000, mds=1000, t=search_types[search_index], ns=200, r=2, k=k, nw=5, wf=500, dr=7))
print(data)
if 0==show_type: runtimes.append(data.timeTotal)
elif 1==show_type: runtimes.append(data.numTQSP)
elif 2==show_type: runtimes.append(data.numAccessedRTreeNode)
chart.draw_line(runtimes, type)
chart.show()
def loadCsv(dir, rFanout=50, alpha=0.5, steepD=0.1, h=15, om=1, oe='1.0E-4', ns=50, t=1, k=5000, nw=2, mpts=5,
eps='0.001', xi='0.001', maxPNeiByte=2147483631):
fp = PathUtility.sample_res_path(dir, rFanout, alpha, steepD, h, om, oe, ns, t, k, nw, mpts,eps, xi, maxPNeiByte)
filename = fp[fp.rindex('res') + 4:]
alldata = []
reader = IterableReader(fp)
index = 0
for line in reader:
if index == 0:
index += 1
continue
strArr = line.split(',')
data = Data(filename)
data.numSample = 1
data.numRangeRtree = int(strArr[7])
data.numOpticFastRange = int(strArr[16])
data.numOpticLuceneRange = int(strArr[18])
data.timeTotal = int(strArr[24])
data.numCluster = int(strArr[25])
alldata.append(data)
return alldata
def draw_word_frequency(title='WORD_FREQUENCY', base_y=None, rotation=0, fpath='test.pdf'):
if base_y!=None: ys = [base_y+i*100 for i in range(0, 14)]
xs=[i for i in range(0, 10)]
x_txts = [0, r'$5\times10^1$', r'$1\times10^2$',
r'$5\times10^2$', r'$1\times10^3$',
r'$5\times10^3$', r'$1\times10^4$',
r'$5\times10^4$', r'$1\times10^5$',
r'$1\times10^6$'
]
if base_y!=None: chart = LineChart(xs, x_txts, ys=ys, yscale='liner', xLabel=title, yLabel='Runtime (ms)', title=title, xlabel_rotation=rotation, fpath = fpath)
else: chart = LineChart(xs, x_txts, xLabel=title, yLabel='Runtime (ms)', title=title, xlabel_rotation=rotation, fpath = fpath)
# chart.ax.grid('on')
wfs = [0, 50, 100, 500, 1000, 5000, 10000, 50000, 100000, 1000000]
base_dir = 'D:\\nowMask\KnowledgeBase\\sample_result\\yago2s_single_date\\word_fequency_test\\'
runtimes = []
for wf in wfs:
data = Data.getData(fp=PathUtility.sample_res_path(base_dir, wf=wf))
runtimes.append(data.timeTotal)
chart.draw_line(runtimes, r'$SPTD^*$')
chart.show()