def __init__(self,word):
self.word = word
self.freq_word = 0#freq of the words in document
self.freq_together = 0# frq of that word occuring with that keyword
self.doc_freq_obj = doc_freq_class.context()
def get_words_from_proximity(self,keywords,text): #think of how to get nouns from sentence only... !!
#create object of doc_frequency
doc_freq_obj = doc_freq_class.context()
tokens = nltk.word_tokenize(text)
#print "tokens:"
#print tokens
for i in tokens:
if i.isalnum()== False:
tokens.remove(i)
c = nltk.ConcordanceIndex(tokens, key = lambda s: s.lower())
tokens_pos = nltk.pos_tag(tokens)
i = 5
doc_freq = []
df_cnt = 0
print "keywords going to loop",
print keywords
for kw in keywords:
print "keyword::::::::",
print kw
#split keyword not required as kw is list of strings
#k = nltk.word_tokenize(kw)
#print k
#print "keywords in for ",
#print kw
first_word = kw[0] #1st word in keyword
#print "first word"
#print first_word
keyword_len = len(kw)
#print "LEN="+str(keyword_len)
i = 5
nomatch = 0
#print "IN KWD LOOP."
print "offset",
print c.offsets(first_word)
for offset in c.offsets(first_word):
print kw
j = 1
i = 5
#print "Keyword=",
#print kw,
#print " OFFSET=" + str(offset)
nomatch = 0
while j < keyword_len:
#print "in while"
#print tokens[offset+j]
#print kw[j]
if tokens[offset+j].lower() <> kw[j].lower():
#print tokens[offset+j]
#print k[j]
nomatch = 1
break
j = j + 1
if nomatch == 0:
doc_freq.append(document_frequency(kw))
#print "matched kwd",
#print tokens[offset:offset+j-1]
#print tokens[offset-5:offset+5]
i = 5
while i > 0 :
if (offset-i) < 0:
break
if (tokens_pos[offset-i][1] in ["NN","NNP"]) and (tokens_pos[offset-i][1].lower() not in nltk.corpus.stopwords.words('english')):
#doc_freq_obj.get_together_DF("")
#print "dfcnt:" + str(df_cnt)
#print "i: " + str(i)
doc_freq[df_cnt].addneighbour(tokens_pos[offset-i][0])
print tokens_pos[offset-i][0],
#pass
i = i - 1
print "\m/ ",
print kw,
print "\m/ ",
i = 1
while i < 5 :
if (offset+i+(keyword_len-1)) >= len(tokens):
break
if (tokens_pos[offset+i+(keyword_len-1)][1] in ["NN","NNP"]) and (tokens_pos[offset+i+(keyword_len-1)][1].lower() not in nltk.corpus.stopwords.words('english')):
#pass
doc_freq[df_cnt].addneighbour(tokens[offset+i+(keyword_len-1)])
print tokens_pos[offset+i+(keyword_len-1)][0],
i = i + 1
k = 0
print "\n\n"
while k < doc_freq[df_cnt].cnt:
#doc_freq[df_cnt].neighbours[k].freq_word = fd1[context_vectors[CV_cnt].keyword]
doc_freq[df_cnt].neighbours[k].find_doc_freq(doc_freq[df_cnt].keyword)
k = k + 1
doc_freq[df_cnt].neighbours.sort(key=lambda x: x.freq_together, reverse=True)
if doc_freq[df_cnt].cnt > 5:
doc_freq[df_cnt].neighbours = doc_freq[df_cnt].neighbours[:5] #take 10 neighbours with highest weight
doc_freq[df_cnt].cnt = 5
k = 0
#while k < doc_freq[df_cnt].cnt:
print "keyword: ",
for l in doc_freq[df_cnt].keyword:
print l,
print "\n"
print "neighbours: ",
for m in doc_freq[df_cnt].neighbours:
print m.word,
print "\n"
#k += 1
df_cnt = df_cnt + 1
results = search_web(doc_freq)
return results
def get_words_from_proximity(
self, keyword_list,
text): #think of how to get nouns from sentence only... !!
#create object of doc_frequency
doc_freq_obj = doc_freq_class.context()
tokens = nltk.word_tokenize(text)
#print "tokens:"
#print tokens
for i in tokens:
if i.isalnum() == False:
tokens.remove(i)
c = nltk.ConcordanceIndex(tokens, key=lambda s: s.lower())
tokens_pos = nltk.pos_tag(tokens)
i = 5
doc_freq = []
df_cnt = 0
print "keywords going to loop",
print keyword_list
keywords = []
for k in keyword_list:
kw = nltk.word_tokenize(k)
keywords.append(kw)
print "keywords"
print keywords
for kw in keywords:
print "keyword::::::::",
print kw
#split keyword not required as kw is list of strings
#k = nltk.word_tokenize(kw)
#print k
#print "keywords in for ",
#print kw
first_word = kw[0] #1st word in keyword
#print "first word"
#print first_word
keyword_len = len(kw)
#print "LEN="+str(keyword_len)
i = 5
nomatch = 0
#print "IN KWD LOOP."
print "offset",
print c.offsets(first_word)
doc_freq.append(document_frequency(kw))
no_of_times = 0
for offset in c.offsets(first_word):
print kw
j = 1
i = 5
#print "Keyword=",
#print kw,
#print " OFFSET=" + str(offset)
nomatch = 0
while j < keyword_len:
#print "in while"
#print tokens[offset+j]
#print kw[j]
if tokens[offset + j].lower() <> kw[j].lower():
#print tokens[offset+j]
#print k[j]
nomatch = 1
break
j = j + 1
if nomatch == 0:
#print "matched kwd",
#print tokens[offset:offset+j-1]
#print tokens[offset-5:offset+5]
i = 5
while i > 0:
if (offset - i) < 0:
break
if (tokens_pos[offset - i][1] in [
"NN", "NNP"
]) and (tokens_pos[offset - i][1].lower()
not in nltk.corpus.stopwords.words('english')):
#doc_freq_obj.get_together_DF("")
#print "dfcnt:" + str(df_cnt)
#print "i: " + str(i)
doc_freq[df_cnt].addneighbour(tokens_pos[offset -
i][0])
print tokens_pos[offset - i][0],
#pass
i = i - 1
print "\m/ ",
print kw,
print "\m/ ",
i = 1
while i < 5:
if (offset + i + (keyword_len - 1)) >= len(tokens):
break
if (tokens_pos[offset + i + (keyword_len - 1)][1] in [
"NN", "NNP"
]) and (tokens_pos[offset + i +
(keyword_len - 1)][1].lower()
not in nltk.corpus.stopwords.words('english')):
#pass
doc_freq[df_cnt].addneighbour(
tokens[offset + i + (keyword_len - 1)])
print tokens_pos[offset + i +
(keyword_len - 1)][0],
i = i + 1
k = 0
print "\n\n"
while k < doc_freq[df_cnt].cnt:
#doc_freq[df_cnt].neighbours[k].freq_word = fd1[context_vectors[CV_cnt].keyword]
doc_freq[df_cnt].neighbours[k].find_doc_freq(
doc_freq[df_cnt].keyword)
k = k + 1
doc_freq[df_cnt].neighbours.sort(
key=lambda x: x.freq_together, reverse=True)
if doc_freq[df_cnt].cnt > 5:
doc_freq[df_cnt].neighbours = doc_freq[
df_cnt].neighbours[:
5] #take 10 neighbours with highest weight
doc_freq[df_cnt].cnt = 5
k = 0
#while k < doc_freq[df_cnt].cnt:
print "keyword: ",
for l in doc_freq[df_cnt].keyword:
print l,
print "\n"
print "neighbours: ",
for m in doc_freq[df_cnt].neighbours:
print m.word,
print "\n"
#k += 1
no_of_times = no_of_times + 1
if no_of_times >= 2:
break
#import pdb;pdb.set_trace();
df_cnt = df_cnt + 1
#results = search_web(doc_freq)
print doc_freq
return doc_freq
def __init__(self, word):
self.word = word
self.freq_word = 0 #freq of the words in document
self.freq_together = 0 # frq of that word occuring with that keyword
self.doc_freq_obj = doc_freq_class.context()
def main():
obj = proper_noun()
text = obj.scrape(sys.argv[1])
title = alchemyObj.URLGetTitle(sys.argv[1])
soup = BeautifulSoup(title)
raw = soup('title')
tokens_title_first = [str(title.text) for title in raw]
#tokens_title = ['Three', 'Musketeers']
print "title::",
print tokens_title_first
#text = original
### Take nouns in title
tokens_title_first = str(tokens_title_first[0])
print tokens_title_first
tokens_title_temp = nltk.word_tokenize(tokens_title_first)
tokens_title_pos = nltk.pos_tag(tokens_title_temp)
print "tokens_title_temp::",
print tokens_title_temp
tokens_title = [] ##create duplicate list
for t in tokens_title_temp:
index = tokens_title_temp.index(t)
print "t::" + t
print "index::" + str(index)
print "tag::" + tokens_title_pos[index][1]
print "len" + str(len(t))
if (t.isalpha() and (tokens_title_pos[index][1] == "NNP") and (len(t) >= 3)):
# tokens_title.remove(t)
tokens_title.append(t)
tokens_title.sort()
tokens_title = list(tokens_title for tokens_title,_ in itertools.groupby(tokens_title))
print "title::",
print tokens_title
list_of_NNPs = obj.get_nnp_ngrams(text,5,0)
#list_of_NNPs = [['Three','Musketeers'],['Alexandre', 'Dumas']]#,['Cardinal', 'Richelieu'],['Athos'],['Although'],['Porthos'] ]
print "list of NNPs: ",
print list_of_NNPs
if len(list_of_NNPs)>3: ######
list_of_NNPs = list_of_NNPs[0:3] ########
doc_freq_obj = doc_freq_class.context()
print "getting doc freq"
max_df = []
for n in list_of_NNPs:
print "got n"
max_freq = 0
for t in tokens_title:
print "got t"
df = doc_freq_obj.get_together_DF(n,t)
if df > max_freq:
max_freq = df
print "ngram:",
print n
print "title word:",
print t
print "df:",
print df
max_df.append(max_freq)
i = 0
for df in max_df:
for i in range(len(max_df)-1):
if max_df[i]<max_df[i+1]:
t = list_of_NNPs[i]
list_of_NNPs[i]=list_of_NNPs[i+1]
list_of_NNPs[i+1]= t
t1 = max_df[i]
max_df[i]=max_df[i+1]
max_df[i+1] = t1
#i = 0
for i in range(len(list_of_NNPs)):
print "keyword: ",
print list_of_NNPs[i]
print "df:",
print max_df[i]
if len(list_of_NNPs)>3:
list_of_NNPs = list_of_NNPs[0:3]#*********
#list_of_NNPs.sort()
#list_of_NNPs_final = list(list_NNPs for list_NNPs,_ in itertools.groupby(list_of_NNPs))
#list_of_NNPs_final.sort()
print "\n\nfinal list:",
print list_of_NNPs
nearbywordsObj = getnearbywords_intokens.getnearbywords()
nearbywordsObj.get_words_from_proximity(list_of_NNPs,text)
def keywords(self, url, text):
# import pdb;pdb.set_trace();
title = alchemyObj.URLGetTitle(url)
soup = BeautifulSoup(title)
raw = soup('title')
tokens_title_first = [str(title.text) for title in raw]
print "title::",
print tokens_title_first
#text = original
### Take nouns in title
tokens_title_first = str(tokens_title_first[0])
print tokens_title_first
tokens_title_temp = nltk.word_tokenize(tokens_title_first)
tokens_title_pos = nltk.pos_tag(tokens_title_temp)
print "tokens_title_temp::",
print tokens_title_temp
tokens_title = [] ##create duplicate list
for t in tokens_title_temp:
index = tokens_title_temp.index(t)
print "t::" + t
print "index::" + str(index)
print "tag::" + tokens_title_pos[index][1]
print "len" + str(len(t))
if (t.isalpha() and (tokens_title_pos[index][1] == "NNP")
and (len(t) >= 3)):
# tokens_title.remove(t)
tokens_title.append(t)
tokens_title.sort()
tokens_title = list(
tokens_title
for tokens_title, _ in itertools.groupby(tokens_title))
print "title::",
print tokens_title
list_of_NNPs = self.get_nnp_ngrams(text, 5, 0)
print "list of NNPs: ",
print list_of_NNPs
doc_freq_obj = doc_freq_class.context()
print "getting doc freq"
max_df = []
# for n in list_of_NNPs:
#print "got n"
# max_freq = 0
# for t in tokens_title:
# print "got t"
# df = doc_freq_obj.get_together_DF(n,t)
# if df > max_freq:
# max_freq = df
# print "ngram:",
# print n
# print "title word:",
# print t
# print "df:",
# print df
# max_df.append(max_freq)
#i = 0
#for df in max_df:
# for i in range(len(max_df)-1):
# if max_df[i]<max_df[i+1]:
# t = list_of_NNPs[i]
# list_of_NNPs[i]=list_of_NNPs[i+1]
# list_of_NNPs[i+1]= t
# t1 = max_df[i]
# max_df[i]=max_df[i+1]
# max_df[i+1] = t1
# for i in range(len(list_of_NNPs)):
# print "keyword: ",
# print list_of_NNPs[i]
# print "df:",
# print max_df[i]
# print "\n\nfinal list:",
# print list_of_NNPs
return list_of_NNPs
def keywords(self,url,text):
# import pdb;pdb.set_trace();
title = alchemyObj.URLGetTitle(url)
soup = BeautifulSoup(title)
raw = soup('title')
tokens_title_first = [str(title.text) for title in raw]
print "title::",
print tokens_title_first
#text = original
### Take nouns in title
tokens_title_first = str(tokens_title_first[0])
print tokens_title_first
tokens_title_temp = nltk.word_tokenize(tokens_title_first)
tokens_title_pos = nltk.pos_tag(tokens_title_temp)
print "tokens_title_temp::",
print tokens_title_temp
tokens_title = [] ##create duplicate list
for t in tokens_title_temp:
index = tokens_title_temp.index(t)
print "t::" + t
print "index::" + str(index)
print "tag::" + tokens_title_pos[index][1]
print "len" + str(len(t))
if (t.isalpha() and (tokens_title_pos[index][1] == "NNP") and (len(t) >= 3)):
# tokens_title.remove(t)
tokens_title.append(t)
tokens_title.sort()
tokens_title = list(tokens_title for tokens_title,_ in itertools.groupby(tokens_title))
print "title::",
print tokens_title
list_of_NNPs = self.get_nnp_ngrams(text,5,0)
print "list of NNPs: ",
print list_of_NNPs
doc_freq_obj = doc_freq_class.context()
print "getting doc freq"
max_df = []
# for n in list_of_NNPs:
#print "got n"
# max_freq = 0
# for t in tokens_title:
# print "got t"
# df = doc_freq_obj.get_together_DF(n,t)
# if df > max_freq:
# max_freq = df
# print "ngram:",
# print n
# print "title word:",
# print t
# print "df:",
# print df
# max_df.append(max_freq)
#i = 0
#for df in max_df:
# for i in range(len(max_df)-1):
# if max_df[i]<max_df[i+1]:
# t = list_of_NNPs[i]
# list_of_NNPs[i]=list_of_NNPs[i+1]
# list_of_NNPs[i+1]= t
# t1 = max_df[i]
# max_df[i]=max_df[i+1]
# max_df[i+1] = t1
# for i in range(len(list_of_NNPs)):
# print "keyword: ",
# print list_of_NNPs[i]
# print "df:",
# print max_df[i]
# print "\n\nfinal list:",
# print list_of_NNPs
return list_of_NNPs
