def findSourceDoc(self):
root.filename = tkinter.filedialog.askopenfilename(
initialdir=os.path.dirname(os.path.realpath(__file__)),
title="Select file to search",
filetypes=(("txt files", "*.txt"), ("all files", "*.*")))
if root.filename == '':
print("cancelled")
else:
srcCandidate, dupl = FindSourceDoc.run(root.filename,
root,
filter=False,
filamt=0)
try:
ParseXML.run(root.filename)
except:
print("already in annotation")
dicc = g.openResult('output/annotation.csv')
annoSource = dicc.get(ntpath.basename(root.filename[:-4]))[0]
annoDup = dicc.get(ntpath.basename(root.filename[:-4]))[1]
p, r, f = g.allmeasure(srcCandidate, dupl, annoSource, annoDup)
print(p)
print(r)
print(f)
result = tkinter.Toplevel(root)
result.minsize(200, 200)
result.title("Evaluation " + ntpath.basename(root.filename))
tkinter.Label(result, text="Precision ").grid(row=0,
column=0,
padx=3)
tkinter.Label(result, text="Recall ").grid(row=1, column=0, padx=3)
tkinter.Label(result, text="F1 Score ").grid(row=2,
column=0,
padx=3)
tkinter.Label(result, text=round(p, 3)).grid(row=0,
column=1,
padx=3)
tkinter.Label(result, text=round(r, 3)).grid(row=1,
column=1,
padx=3)
tkinter.Label(result, text=round(f, 3)).grid(row=2,
column=1,
padx=3)
def simul(filteramt, file):
if file == "Simulated":
file = glob.glob("testdoc/simulatedFiles/*.txt")
else:
file = glob.glob("testdoc/artificialFile/*.txt")
#print(file)
#print(filteramt)
precision = []
recall = []
F1 = []
docname = []
count = 0
for fi in file:
count += 1
docname.append(count)
if fi == '':
print('cancelled')
else:
if filteramt == 0:
srcCandidate, dupl = FindSourceDoc.run(fi,
root,
show=False,
filter=False,
filamt=0)
else:
srcCandidate, dupl = FindSourceDoc.run(fi,
root,
show=False,
filamt=filteramt)
try:
ParseXML.run(fi)
except:
print("already in annotation")
dicc = g.openResult('output/annotation.csv')
annoSource = dicc.get(ntpath.basename(fi[:-4]))[0]
annoDup = dicc.get(ntpath.basename(fi[:-4]))[1]
p, r, f = g.allmeasure(srcCandidate, dupl, annoSource, annoDup)
precision.append(p)
recall.append(r)
F1.append(f)
plt.subplot(2, 2, 1)
plt.bar(docname, precision)
plt.title('Precision = ' + str(round(sum(precision) / len(precision), 3)))
plt.ylim(top=1.05)
plt.subplot(2, 2, 2)
plt.bar(docname, recall)
plt.title('Recall = ' + str(round(sum(recall) / len(recall), 3)))
plt.ylim(top=1.05)
plt.subplot(2, 2, 3)
plt.bar(docname, recall)
plt.title('F1 Score = ' + str(round(sum(F1) / len(F1), 3)))
plt.ylim(top=1.05)
plt.subplots_adjust(hspace=0.3)
plt.show()
