)
x += 1
outfile.write("</edges>\n")
outfile.write("</graph></gexf>\n")
outfile.close()
# Get TFIDF table list
tfidf_tblist = obj_db.getTFIDFTableList()
# Get Segmented table list
seg_tblist = obj_db.getTableList()
print seg_tblist
for i in range(5):
dict_tfidf = {}
results = mysql.queryrows("SELECT * FROM `" + tfidf_tblist[i] + "`")
for row in results:
list_pairs = row[1].split("|")
for pair in list_pairs:
term_tfidf = pair.split("@")
term = term_tfidf[0]
tfidf = float(term_tfidf[1])
if term in dict_tfidf:
continue
else:
dict_tfidf[term] = tfidf
# Remove number and not english word in dictionary
for key, val in dict_tfidf.items():
if verifyEngNum(key) == 1 or verifyEngNum(key) == 3:
dict_tfidf.pop(key)
continue
else:
# Init ang reset feature and category dataset
avg_feature = []
avg_category = []
max_feature = []
max_category = []
# Rest All value to 0.0 in dictionary
dict_avg_zero = dict.fromkeys(dict_flted_avgchi, 0.0)
dict_max_zero = dict.fromkeys(dict_flted_maxchi, 0.0)
sql_getContent = "SELECT `ClsNo1`, `ScoreContent` FROM `" + tfidf_viewlist[view_num] + "`"
# Get All result
view_result = mysql.queryrows(sql_getContent)
for view_row in view_result:
view_content = view_row[1]
term_tfidf_list = view_content.split("|")
avg_tfidf_score_vactor = getTFIDFScoreVector(dict_avg_zero, term_tfidf_list)
max_tfidf_score_vextor = getTFIDFScoreVector(dict_max_zero, term_tfidf_list)
avg_feature.append(avg_tfidf_score_vactor)
avg_category.append(int(view_row[0]))
max_feature.append(max_tfidf_score_vextor)
max_category.append(int(view_row[0]))
print 'AVG Feature dataset: ' + str(len(avg_feature))
print 'AVG Category dataset: ' + str(len(avg_category))
print 'MAX Feature dataset: ' + str(len(max_feature))
def pairReliablility(user1, user2, coder_clsfi):
len_n = len(coder_clsfi)
M = 0
for key, value in coder_clsfi.items():
if value[user1] == value[user2]:
M += 1
rebilty = (2.0 * M) / (2.0 * len_n)
print "User: "******", " + str(user2)
print "M (Number of totally agreement): " + str(M)
print "N1,N2 (Should agree with number): " + str(len_n)
print "Mutual consent degree = 2M/(N1+N2): " + str(rebilty) + "\n\n"
sql_get = "SELECT `SamplingNo`, `ClsNo1`, `UserId` FROM `CoderCompare` ORDER BY `CoderCompare`.`SamplingNo` ASC"
result = mysql.queryrows(sql_get)
coder_clsfi = {}
for row in result:
sno = int(row[0])
cls = int(row[1])
user = int(row[2])
if sno in coder_clsfi:
coder_clsfi[sno][user] = cls
else:
coder_clsfi[sno] = {}
coder_clsfi[sno][user] = cls
print coder_clsfi
