def process_query(queryt, dictionaryword, dictionarytitleq, dictionaryc):
querydt = queryt.split(',')
querytt = None
query = None
queryc = None
for i in querydt:
if 'T:' in i:
querytt = i.split(':')
querytt = querytt[1]
elif 'B:' in i:
query = i.split(':')
query = query[1]
elif 'C:' in i:
queryc = i.split(':')
queryc = queryc[1]
else:
query = i
querytt = i
if query is not None:
queryterms = re.split("[^a-zA-Z]+", query)
for word in queryterms:
word = word.lower()
if word not in stopword.stopword:
word = stem(word)
if word not in dictionaryword:
dictionaryword[word] = 1
else:
dictionaryword[word] += 1
if querytt is not None:
queryterms = re.split("[^a-zA-Z]+", querytt)
for word in queryterms:
word = word.lower()
if word not in stopword.stopword:
word = stem(word)
if word not in dictionarytitleq:
dictionarytitleq[word] = 1
else:
dictionarytitleq[word] += 1
if queryc is not None:
queryterms = re.split("[^a-zA-Z]+", queryc)
for word in queryterms:
word = word.lower()
if word not in stopword.stopword:
word = stem(word)
if word not in dictionarytitleq:
dictionaryc[word] = 1
else:
dictionaryc[word] += 1
def createDocAndTokenHashes_WithStemming():
print("Creating Document Hash and Token Hash with stemming....")
files = os.listdir("C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\ap89_collection")
doc_id=1
token_increment = 1
doc_hash = dict()
token_hash = dict()
for file in files:
with open("C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\ap89_collection\\"+file) as f:
doc_file = f.read()
docs=re.findall(r'<DOC>(.*?)</DOC>',doc_file,re.DOTALL)
if docs:
for doc in docs:
li_doc_no=re.findall(r'<DOCNO>(.*?)</DOCNO>',doc)
li_texts=re.findall(r'<TEXT>(.*?)</TEXT>',doc,re.DOTALL)
doc_no= ''.join(map(str, li_doc_no))
texts=''.join(map(str, li_texts))
dlen=0
for m in re.finditer(r'\w+(\.?\w+)*', texts.lower()):
token_noStem = m.group(0)
token= stem(token_noStem)
dlen+=1
if token not in token_hash.keys():
token_hash[token] = token_increment
token_increment+=1
doc_hash[doc_no]=(doc_id,dlen)
doc_id+=1
print("Docs= "+str(doc_id-1))
print("Tokens with stemming= "+str(token_increment-1))
pickle.dump( doc_hash, open( "C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\HW2_Data\\Doc_Hash_withStemming", "wb" ) )
pickle.dump( token_hash, open( "C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\HW2_Data\\Token_Hash_withStemming", "wb" ) )
def get_feature_list():
feature_list = set()
for line in open("./../data/featurelist.txt"):
word = line.strip()
feature_list.add(stem(word.lower()))
return feature_list
def text_processing(texts):
if isinstance(texts, list):
clean = []
for st in texts:
clean.extend([
stem(w) for w in st.casefold().translate(translator).split()
if w not in stop_words
])
return clean
clean = [
stem(w) for w in texts.casefold().translate(translator).split()
if w not in stop_words
]
return clean
def getP(self, review):
words = re.split('[^A-Za-z0-9_\'\"?!]+',review)
words = filter(None,words)
words = [stem(word.lower()) for word in words]
words = set(words)
pos = (0.0+self.pos_doc)/(self.pos_doc+self.neg_doc)
for word in words:
if len(word) == 1 and word != '?' and word != '!':
continue
if word in self.positive:
pos *= self.positive[word]
else:
pos *= (1.0/self.pos_doc)
neg = (0.0+self.neg_doc)/(self.pos_doc+self.neg_doc)
for word in words:
if len(word) == 1 and word != '?' and word != '!':
continue
if word in self.negative:
neg *= self.negative[word]
else:
neg *= (1.0/self.neg_doc)
print "pos:"+str(pos)
print "neg:"+str(neg)
if pos > neg:
print 'pos'
return 1
else:
print 'neg'
return 0
def read(self, file_name, timelength):
with open(file_name, 'rb') as f:
data = json.load(f)
vocabulary = {}
tempLine = []
lineno = 0
for line in data:
temp = {}
cmnt = re.split(r'[^A-Za-z]+', line["message"])
words = []
for word in cmnt:
word = word.lower()
if word not in self.stopWords:
word = stem(word)
words.append(word)
if word not in vocabulary:
vocabulary[word] = 0
if len(words) != 0:
temp["time"] = line["time"]
temp["lineno"] = lineno
temp["text"] = words
tempLine.append(temp)
lineno += 1
lines = sorted(tempLine, key=lambda e: (e.__getitem__('time')))
self.store(lines, timelength)
return lines, timelength, vocabulary
def getP(self, review):
words = re.split('[^A-Za-z0-9_\']+',review)
words = filter(None,words)
words = [stem(word.lower()) for word in words]
words = set(words)
pos = (0.0+self.pos_count)/(self.pos_count+self.neg_count)
for word in words:
if word in self.positive:
pos *= self.positive[word]
else:
pos *= (1.0/self.pos_count)
neg = (0.0+self.neg_count)/(self.pos_count+self.neg_count)
for word in words:
if word in self.negative:
neg *= self.negative[word]
else:
neg *= (1.0/self.neg_count)
print "pos:"+str(pos)
print "neg:"+str(neg)
if pos > neg:
return 1
else:
return 0
def build_queries():
file = "query_desc.51-100.short.txt"
with open("C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\"+file) as f:
queries = f.readlines()
q_dict = dict()
for query in queries:
new_query = query.split(' ')
li_query_terms=[]
for m in re.finditer(r'\w+(\.?\w+)*', new_query[1].lower()):
li_query_terms.append(m.group(0))
edit_query = ' '.join(map(str, li_query_terms[3:]))
q_dict[new_query[0][:-1]] = edit_query
stop_file = open("C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\stoplist.txt",'r')
stop_words = []
for line in stop_file:
line=line[:-1]
stop_words.append(line)
fin_query=''
query_dict=dict()
for key in q_dict.keys():
fin_query=' '.join([stem(word) for word in q_dict[key].split() if word not in stop_words])
query_dict[key]=fin_query
return query_dict
def build_queries():
file = "query_desc.51-100.short.txt"
with open("C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\"+file) as f:
queries = f.readlines()
q_dict = dict()
for query in queries:
new_query = query.split('.')
new_query.pop()
if len(new_query) > 0 and new_query[0] not in q_dict:
edit_query=new_query[1].split()
edit_query = edit_query[3:]
edit_query = ' '.join(edit_query)
edit_query = edit_query.replace(',','')
edit_query = edit_query.replace('"','')
edit_query = edit_query.replace('-',' ')
q_dict[new_query[0]] = edit_query
stop_fname = "stoplist.txt"
stop_file = open("C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\"+stop_fname,'r')
stop_words = []
for line in stop_file:
line=line[:-1]
stop_words.append(line)
fin_query=''
query_dict=dict()
for key in q_dict.keys():
fin_query=' '.join([stem(word) for word in q_dict[key].split() if word not in stop_words])
query_dict[key]=fin_query
return query_dict
def ans_52():
with open('nlp_51.txt', mode='r') as f:
words = f.read()
words = words.split('\n')
result = '\n'.join(
[word + '\t' + stem(word) for word in words if len(words) != 0])
# list(map(lambda x: [x+'\t'+stem(x)], words))
print(result[:300])
def wordCleanUp(word, irrDict=None):
"""
stem and change verbs to present tense
"""
w = stem(word.lower())
try:
return irrDict[w]
except (TypeError, KeyError):
return w
def read_csv_file(self, filename):
csv_file = csv.DictReader(open(filename,'rb'), delimiter=',',quotechar='"')
self.positive = dict()
#self.pos_alpha = dict()
self.negative = dict()
#self.neg_alpha = dict()
self.pos_doc = 0.0
self.neg_doc = 0.0
for line in csv_file:
review = line['Review']
category =line['Category']
words = re.split('[^A-Za-z0-9_\'\"?!]+',review)
words = filter(None,words)
words = [stem(word.lower()) for word in words]
words = set(words)
if category == '1':
self.pos_doc +=1.0
temp_words = set()
for word in words:
if word == '\'' or word == '\"':
continue
temp_words.add(word)
for word in temp_words:
if self.positive.has_key(word):
self.positive[word] +=1
else:
self.positive[word] = 1
elif category == '0':
self.neg_doc += 1.0
temp_words = set()
for word in words:
if word == '\'' or word == '\"':
continue
temp_words.add(word)
for word in temp_words:
if self.negative.has_key(word):
self.negative[word] += 1
else:
self.negative[word] = 1
#inter_words = set(self.positive.keys())&set(self.negative.keys())
for word in self.positive.keys():
if len(word) == 1 and word != '?' and word != '!':
del self.positive[word]
continue
for word in self.negative.keys():
if len(word) == 1 and word != '?' and word != '!':
del self.negative[word]
continue
for word in self.positive.keys():
self.positive[word] = self.positive[word]/self.pos_doc
for word in self.negative.keys():
self.negative[word] = self.negative[word]/self.neg_doc
def stemm_word(word):
"""
Use Porter stemmer to stem words.
:param word: String
:return: Stemmed word
"""
return stem(word)
def build_table():
table = defaultdict(Counter)
csvs = [f for f in os.listdir('.') if f.endswith('.csv')]
for fname in csvs:
with open(fname, 'rb') as f:
reader = csv.reader(f)
for username, tweet in reader:
for s in (stem(t) for t in tokenize(tweet)):
table[username][s] += 1
return table
def tokenize(message):
"""
Finds all the words in message and returns them as a
unique set
"""
message = message.lower()
all_words = re.findall("[a-z0-9']+", message)
words = [stem(word) for word in all_words]
return set(words)
def build_table():
table = defaultdict(Counter)
dirs = [fname for fname in os.listdir('.') if fname.endswith('.csv')]
for fname in dirs:
with open(fname, 'r') as f:
reader = csv.reader(f)
for username, tweet in reader:
for token in tokenize(tweet.lower()):
if token not in stopwords:
table[username.lower()][stem(token)] += 1
return table
def proc_data(data):
for (i, desc) in data:
if pd.isnull(desc): desc = 'none'
desc = desc.translate(trans_mask).lower().split(' ')
ln_list = []
for word, cnt in Counter(desc).iteritems():
word = word.strip()
word = stem(word)
if word in skip: continue
if word == '': continue
ln_list.append("%i:%i" % (vocab[word], cnt))
yield str(len(ln_list)) + ' ' + ' '.join(ln_list) + '\n'
def getIrrDict(path=".\wordlists\irr.txt"):
'''
irregular verbs stemming
'''
irr = dict()
with open(path, 'r') as f:
for l in f:
w0 = l.split()[0].lower()
for w in l.strip().split():
w1 = stem(w.lower())
if w1 != w0 : irr[w1]=w0
return irr
def calculate_cosine_similarity(alert_type): cos_numerator_sum = 0 cos_denominator_local_count = 0 cos_denominator_news_count = 0 for local_word in alert_type: for news_word in getKeywords(article): if stem(local_word) == stem(news_word[0]): cos_numerator_sum = cos_numerator_sum + news_word[1] cos_denominator_local_count+=1 cos_denominator_news_count_temp = news_word[1]*news_word[1] cos_denominator_news_count = cos_denominator_news_count + cos_denominator_news_count_temp cos_denominator_sum = math.sqrt(cos_denominator_news_count) * math.sqrt(cos_denominator_local_count) cos_similarity = 0 if cos_denominator_sum != 0: cos_similarity = cos_numerator_sum / cos_denominator_sum return cos_similarity
def extract_features(data, dict_features):
data_one_x = np.zeros(len(dict_features) + 1, dtype=np.float64)
data_one_x[0] = 1
for word in data.split(' '):
word = word.strip()
if is_stop_word(word):
continue
word = porter.stem(word)
try:
data_one_x[dict_features[word]] = 1
except:
pass
return data_one_x
def preprocess(text):
'''
preprocesses the given text before feeding it to the classification
system
'''
p = re.compile(r'<script.*?</script>', re.DOTALL)
text = p.sub('', text)
p = re.compile(r'<a href.*?</a>', re.DOTALL)
#TODO use some better name for this
text = p.sub('hyperlinkk', text)
text = remove_spaces(remove_tags(text))
text = text.lower()
p = re.compile(r'[^a-z\s]', re.DOTALL)
text = p.sub('', text)
stemmed_text = ''
for word in text.split():
stemmed_text += stem(word) + " "
return stemmed_text
def addToIndex(wordList, docID, t):
'''
Removes all the non-ASCII words and then performs stemming and then adds in the index at appropriate location.
'''
for word in wordList:
word = word.strip().encode('utf-8')
if word.isalpha() and len(word) > 3 and word not in stopWords:
# Stemming the Words
word = stem(word)
if word not in stopWords:
if word in invertedIndex:
if docID in invertedIndex[word]:
if t in invertedIndex[word][docID]:
invertedIndex[word][docID][t] += 1
else:
invertedIndex[word][docID][t] = 1
else:
invertedIndex[word][docID] = {t: 1}
else:
invertedIndex[word] = dict({docID: {t: 1}})
def read_csv_file(self, filename):
csv_file = csv.DictReader(open(filename,'rb'), delimiter=',',quotechar='"')
self.positive = dict()
#self.pos_alpha = dict()
self.negative = dict()
#self.neg_alpha = dict()
self.pos_count = 0.0
self.neg_count = 0.0
for line in csv_file:
review = line['Review']
category =line['Category']
words = re.split('[^A-Za-z0-9_\']+',review)
words = filter(None,words)
words = [stem(word.lower()) for word in words]
words = set(words)
if category == '1':
self.pos_count +=1.0
for word in words:
if word == '\'':
continue
if self.positive.has_key(word):
self.positive[word] += 1
else:
self.positive[word] = 1
elif category == '0':
self.neg_count += 1.0
for word in words:
if word == '\'':
continue
if self.negative.has_key(word):
self.negative[word] += 1
else:
self.negative[word] = 1
inter_words = set(self.positive.keys())&set(self.negative.keys())
for word in self.positive.keys():
self.positive[word] = self.positive[word]/self.pos_count
for word in self.negative.keys():
self.negative[word] = self.negative[word]/self.neg_count
def stemming(paragraph): temp=split(paragraph) for i in range(len(temp)): temp[i]=stem(temp[i]) temp_x = ' '.join(temp) return temp_x
import string
import urllib
from stemming.porter import stem
news = urllib.urlopen('https://ceiba.ntu.edu.tw/course/35d27d/content/28.txt')
str = news.read()
# lowercase
str = str.lower()
# tokenization
str = str.translate(string.maketrans("",""),string.punctuation)
list = str.split()
# Porter's stemmer
stemlist = []
for x in list:
stemlist.append(stem(x))
# stopword removal
page = urllib.urlopen('http://ir.dcs.gla.ac.uk/resources/linguistic_utils/stop_words')
stopwords_list = page.read().split()
result = [x for x in stemlist if x not in stopwords_list]
# write to file
outstring = " ".join(result)
output = open("output1.txt", "wb+")
output.write(outstring)
output.close()
def query_search(queries, secondary_index, number_of_files):
ranking = {}
queries[3] = re.sub(r"[-.]", '', str(queries[3]))
queries[2] = re.sub(r"[-.]", '', str(queries[2]))
queries[1] = re.sub(r"[-.]", '', str(queries[1]))
queries[0] = re.sub(r"[-.]", '', str(queries[0]))
queryb = re.split('[^a-zA-Z0-9]', str(queries[3]))
queryc = re.split('[^a-zA-Z0-9]', str(queries[2]))
queryi = re.split('[^a-zA-Z0-9]', str(queries[1]))
queryt = re.split('[^a-zA-Z0-9]', str(queries[0]))
if (queryb):
for word in queryb:
if (word == ''):
continue
temp = word.lower()
word = stem(word) + ";p"
rank = None
#print word
rank = findi(word, secondary_index, number_of_files)
for row in rank:
if (row in ranking):
ranking[int(row)] += rank[int(row)]
else:
ranking[int(row)] = rank[int(row)]
leng = {}
word = stem(temp) + ";t"
rank = {}
rank = findi(word, secondary_index, number_of_files)
for row in rank:
if (row in ranking):
ranking[int(row)] += log10(30) * rank[int(row)]
else:
ranking[int(row)] = log10(30) * rank[int(row)]
word = stem(temp) + ";i"
rank = {}
rank = findi(word, secondary_index, number_of_files)
for row in rank:
if (row in ranking):
ranking[int(row)] += rank[int(row)]
else:
ranking[int(row)] = rank[int(row)]
if (queryc):
for word in queryc:
if (word == ''):
continue
word = stem(word.lower()) + ";t"
leng = findi(word, secondary_index, number_of_files)
for row in leng:
if (row in ranking):
ranking[row] += leng[row]
else:
ranking[row] = leng[row]
if (queryt):
for word in queryt:
if (word == ''):
continue
word = stem(word.lower()) + ";t"
leng = findi(word, secondary_index, number_of_files)
for row in leng:
if (row in ranking):
ranking[row] += leng[row]
else:
ranking[row] = leng[row]
if (queryi):
for word in queryi:
if (word == ''):
continue
word = stem(word.lower()) + ";i"
#print word
leng = findi(word, secondary_index, number_of_files)
for row in leng:
if (row in ranking):
ranking[row] += leng[row]
else:
ranking[row] = leng[row]
return ranking
def createPartialIndexes_WithStemnStop():
doc_itr=0
catalog={}
file=open("C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\Partial_Indexer_withStemnStop","a")
file.close()
doc_hash = pickle.load( open( "C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\HW2_Data\\Doc_Hash_withStemnStop", "rb" ) )
token_hash = pickle.load( open( "C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\HW2_Data\\Token_Hash_withStemnStop", "rb" ) )
files = os.listdir("C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\ap89_collection")
dblock_hash={}
stop_file = open("C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\stoplist.txt",'r')
stop_words = []
for line in stop_file:
line=line[:-1]
stop_words.append(line)
for file in files:
with open("C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\ap89_collection\\"+file) as f:
doc_file = f.read()
docs=re.findall(r'<DOC>(.*?)</DOC>',doc_file,re.DOTALL)
if docs:
for doc in docs:
if (doc_itr%1000 == 0 and doc_itr!=0):
#Dump to index
indexfile_content = ''
file=open("C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\HW2_Data\\Partial_Indexer_withStemnStop",'a')
indexfile_offset = file.tell()
file.close()
for token_id in dblock_hash.keys():
token_content = str(token_id)+"{"
for doc_id in dblock_hash[token_id].keys():
token_content += str(doc_id)+": ("+ dblock_hash[token_id][doc_id] + ");"
token_content += "}"
if token_id in catalog.keys():
tkn_ent_list = catalog[token_id]
tkn_ent_list.append((str(indexfile_offset)+","+str(len(token_content))))
catalog[token_id] = tkn_ent_list
else:
catalog[token_id]=[(str(indexfile_offset)+","+str(len(token_content)))]
indexfile_content += token_content
indexfile_offset+=len(token_content)
file=open("C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\HW2_Data\\Partial_Indexer_withStemnStop",'a')
file.write(indexfile_content)
file.close()
dblock_hash={}
print("Dumped " +str(doc_itr)+ " Documents after Stemming and with Stop Words removed")
li_doc_no=re.findall(r'<DOCNO>(.*?)</DOCNO>',doc)
li_texts=re.findall(r'<TEXT>(.*?)</TEXT>',doc,re.DOTALL)
doc_no= ''.join(map(str, li_doc_no))
doc_id = doc_hash[doc_no][0]
texts=''.join(map(str, li_texts))
texts=texts.lower()
tkn_cnt=0
for m in re.finditer(r'\w+(\.?\w+)*', texts.lower()):
token_noStem = m.group(0)
if token_noStem not in stop_words:
token= stem(token_noStem)
tkn_cnt+=1
token_id=token_hash[token]
if token_id in dblock_hash.keys():
if doc_id in dblock_hash[token_id].keys():
dblock_hash[token_id][doc_id]+= ","+str(tkn_cnt)
else:
dblock_hash[token_id][doc_id]=str(tkn_cnt)
else:
dblock_hash[token_id]={}
dblock_hash[token_id][doc_id]=str(tkn_cnt)
doc_itr+=1
#Dump the last chunk of dblocks to index
indexfile_content = ''
file=open("C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\HW2_Data\\Partial_Indexer_withStemnStop",'a')
indexfile_offset = file.tell()
file.close()
for token_id in dblock_hash.keys():
token_content = str(token_id)+"{"
for doc_no in dblock_hash[token_id].keys():
token_content += str(doc_no)+": ("+ dblock_hash[token_id][doc_no] + ");"
token_content += "}"
if token_id in catalog.keys():
tkn_ent_list = catalog[token_id]
tkn_ent_list.append((str(indexfile_offset)+","+str(len(token_content))))
catalog[token_id] = tkn_ent_list
else:
catalog[token_id]=[(str(indexfile_offset)+","+str(len(token_content)))]
indexfile_content += token_content
indexfile_offset+=len(token_content)
file=open("C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\HW2_Data\\Partial_Indexer_withStemnStop",'a')
file.write(indexfile_content)
file.close()
pickle.dump( catalog, open( "C:\\Users\\hp\\Desktop\\IR_Documents\\AP_DATA\\HW2_Data\\Dummy_Catalog_withStemnStop", "wb" ) )
print("Dumped All Documents after Stemming and with Stop Words removed")
def stemming(a_list_of_words):
stemmed = [stem(words) for words in a_list_of_words]
return stemmed
def queryNormal(queryWords):
wordsToSearch = list()
for word in queryWords:
word = word.lower().strip()
if word not in stopWords:
word = stem(word)
if word.isalpha() and len(word) >= 3 and word not in stopWords:
wordsToSearch.append(word)
globalSearch = dict(list())
for word in wordsToSearch:
loc = bisect(secondaryIndex, word)
startFlag = False
if loc - 1 >= 0 and secondaryIndex[loc - 1] == word:
startFlag = True
if loc - 1 != 0:
loc -= 1
if loc + 1 == len(secondaryIndex) and secondaryIndex[loc] == word:
loc += 1
primaryFile = "finalIndex/index" + str(loc) + ".txt"
file = open(primaryFile, "r")
data = file.read()
if startFlag:
startIndex = data.find(word + "=")
else:
startIndex = data.find("\n" + word + "=")
endIndex = data.find("\n", startIndex + 1)
reqLine = data[startIndex:endIndex]
pl = reqLine.split("=")[1].split(",")
numDoc = len(pl)
idf = log10(noDocs / numDoc)
for i in pl:
docID, entry = i.split(":")
if docID in globalSearch:
globalSearch[docID].append(entry + "_" + str(idf))
else:
globalSearch[docID] = [entry + "_" + str(idf)]
lengthFreq = dict(dict())
regEx = re.compile(r'(\d+|\s+)')
for k in globalSearch:
weightedFreq = 0
n = len(globalSearch[k])
for x in globalSearch[k]:
x, idf = x.split("_")
x = x.split("#")
for y in x:
lis = regEx.split(y)
tagType, freq = lis[0], lis[1]
if tagType == "t":
weightedFreq += int(freq) * 1000
elif tagType == "i" or tagType == "c" or tagType == "r" or tagType == "e":
weightedFreq += int(freq) * 50
elif tagType == "b":
weightedFreq += int(freq)
if n in lengthFreq:
lengthFreq[n][k] = float(log10(1 + weightedFreq)) * float(idf)
else:
lengthFreq[n] = {k: float(log10(1 + weightedFreq)) * float(idf)}
count = 0
flag = False
# resultList = []
K = 10
for k, v in sorted(lengthFreq.items(), reverse=True):
for k1, v1 in sorted(v.items(), key=itemgetter(1), reverse=True):
print docTitleMap[k1]
count += 1
if count == K:
flag = True
break
if flag:
break
def tokenize(s):
return [stem(w) for w in re.findall('\w+', s.lower()) if w not in stopwords]
create_doc_term()
create_term_info()
avg_doclen = sum_doclen/num_doc
ignore_file = open("ignore_tokens.txt")
ignore_tokens = list(ignore_file.read().split())
query_list = []
xmldoc = minidom.parse('topics.xmls')
itemlist = xmldoc.getElementsByTagName('topic')
for s in itemlist:
qnumber = s.attributes['number'].value
Q = s.getElementsByTagName('query').item(0).childNodes[0].data.lower()
Qwords = Q.split(' ')
stem_words = []
query_str = ''
for w in Qwords:
if w not in ignore_tokens:
stem_word = stem(w)
stem_words.append(stem_word)
query_str = ' '.join(stem_words)
query = [qnumber, query_str]
query_list.append(query)
Rank(query_list)
else:
incorrect_usage()
else:
incorrect_usage()
def stemming(a_list_of_words):
#return a list of words
return [stem(i) for i in a_list_of_words]
def getStemmed (self, word): return stem(word)
def stemming(lst):
stem_lst = []
for wrd in lst:
stem_lst.append(wrd + "\t" + stem(wrd))
return stem_lst
import re
from stemming import porter
def get_sentence():
reg = re.compile(r'(.*?[.;:?!])\s([A-Z].*?\.)')
with open('nlp.txt') as f:
for l in f:
l = l.strip()
match = reg.split(l)
if len(match) > 0:
for line in match:
if len(line) > 0:
yield (line.strip())
if __name__ == '__main__':
for line in get_sentence():
print(''.join([
porter.stem(word.strip(',.')) + '\n' for word in line.split(' ')
]))
def stemming(a_list_of_words):
#return a list of words
return [stem(i) for i in a_list_of_words]
inrevision = False
ns = 0
l_hitext=[]
dicw={}
elif tname == 'revision':
# Do not pick up on revision id's
inrevision = True
elif tname == 'redirect':
dicw={}
rtitle=elem.attrib;
if rtitle is not None:
l_hitext=re.split("[^a-zA-Z]+", rtitle['title'])
for word in l_hitext:
word=word.lower()
if word not in stopword.stopword:
word=stem(word)
if word not in dicw:
dicw[word]=1
else:
dicw[word]+1
for key in dicw:
if key not in invit:
invit[key]=id+'t'+str(dicw[key])+':'
else:
invit[key]=invit[key]+(id+'t'+str(dicw[key])+':')
l_hitext=[]
dicw={}
else:
if tname == 'title':
title = elem.text
elif tname == 'id' and not inrevision:
'either,else,ever,every,for,from,get,got,had,has,have,he,her,hers,'
'him,his,how,however,i,if,in,into,is,it,its,just,least,let,like,'
'likely,may,me,might,most,must,my,neither,no,nor,not,of,off,often,'
'on,only,or,other,our,own,rather,said,say,says,she,should,since,so,'
'some,than,that,the,their,them,then,there,these,they,this,tis,to,too,'
'twas,us,wants,was,we,were,what,when,where,which,while,who,whom,why,'
'will,with,would,yet,you,your').lower().split(',')
def is_stopword(string):
return string.lower() in stop_words
if __name__ == '__main__':
word_counter = Counter()
with open('sentiment.txt') as f:
for line in f:
for word in line[3:].split(' '):
word = word.strip()
if is_stopword(word):
continue
word = porter.stem(word)
if word != '!' and word != '?' and len(word) <= 1:
continue
word_counter.update([word])
features = [word for word, count in word_counter.items() if count >= 6]
with open('features.txt', 'w') as f:
print(*features, sep='\n', file=f)
def main():
for word in sys.stdin:
word = word.strip()
s_word = porter.stem(word)
s2_word = porter2.stem(word)
print "{}\t{}\t{}".format(word, s_word, s2_word)
# wor=list(set(words))
target = [x for x in temp if x not in commonwords]
target = gram_1(target)
length_target = sum(target[i][1] for i in range(len(target)))
'''
sentences =[]
for i in range(1,476):
file=open('C:/Users/prakash chand gupta/Desktop/AI PROJECT/Articles2/'+str(i)+'.txt','r')
p=file.read().replace("'",' ').replace("`",'').replace(";",'').replace(":",'').replace("!",'.').replace('"',"").replace('\xe2',"").replace('\x80',"").replace('\x9d',"").replace('\x91',"").replace('\x92',"").replace('\x93',"").replace('\x94',"").replace('\x95',"").replace('\x96',"").replace('\x97',"").replace('\x98',"").replace('\x99',"").lower()
p=unicodedata.normalize('NFKD', unicode(p,errors='ignore')).encode('ascii','ignore')
temp = split(p)
temp2 = [x for x in temp if x not in commonwords]
for i in range(len(temp2)):
temp2[i]=stem(temp2[i])
temp3 = [x for x in temp2 if x not in commonwords]
temp3 = gram_1(temp3)
sentences.append(temp3)
'''
source_distance=[]
target_distance=[]
'''
'''
for i in range(len(sentences)):
length_temp = sum(sentences[i][j][1] for j in range(len(sentences[i])))
count=0
if len(sentences[i]) < len(source):
for k in range(len(sentences[i])):
for j in range(len(source)):
def _stem(self, term):
return porter.stem(term)
def stemming(lst):
stem_lst = []
for wrd in lst:
stem_lst.append(wrd + "\t" + stem(wrd))
return stem_lst
'C:/Users/prakash chand gupta/Desktop/AI PROJECT/Articles2/' + str(i) +
'.txt', 'r')
p = file.read().replace("'", ' ').replace("`", '').replace(
";", '').replace(":", '').replace("!", '.').replace('"', "").replace(
'\xe2', "").replace('\x80', "").replace('\x9d', "").replace(
'\x91', "").replace('\x92', "").replace('\x93', "").replace(
'\x94',
"").replace('\x95', "").replace('\x96', "").replace(
'\x97', "").replace('\x98', "").replace('\x99',
"").lower()
p = unicodedata.normalize('NFKD', unicode(p, errors='ignore')).encode(
'ascii', 'ignore')
temp = split(p)
temp2 = [x for x in temp if x not in commonwords]
for i in range(len(temp2)):
temp2[i] = stem(temp2[i])
temp3 = [x for x in temp2 if x not in commonwords]
temp3 = gram_1(temp3)
sentences.append(temp3)
'''
source_distance=[]
target_distance=[]
'''
'''
for i in range(len(sentences)):
length_temp = sum(sentences[i][j][1] for j in range(len(sentences[i])))
count=0
if len(sentences[i]) < len(source):
for k in range(len(sentences[i])):
for j in range(len(source)):
if sentences[i][k][0] == source[j][0] and sentences[i][k][0]!='':
def stemming(paragraph):
temp = split(paragraph)
for i in range(len(temp)):
temp[i] = stem(temp[i])
temp_x = ' '.join(temp)
return temp_x
def tokenize(doc, stopwords):
reg = re.compile(r'(\w+(?:\.?\w+)*)')
return map(lambda t: stem(t),
filter(lambda t: t not in stopwords, reg.findall(doc.lower())))
def analyze_term(term, stopwords):
if term in stopwords:
return None
else:
return stem(term.lower())
f_docids = open("docids.txt", "w")
f_termids = open("termids.txt", "w")
f_doc_index = open("doc_index.txt", "w")
f_doc_read = open("doc_index.txt", "r")
for file in os.listdir(path):
docid_str = file + '\t' + str(docid) + '\n'
f_docids.write(docid_str)
count = 1
wordList = tokenize (file)
doc_words = collections.OrderedDict()
stem_list = []
for word in wordList:
stem_word = stem(word[0])
if stem_word not in doc_words.keys() and word[0] not in ignore_tokens:
doc_words[stem_word] = 1
if stem_word not in all_tokens.keys():
all_tokens.update({stem_word : termid})
termid_str = ''.join([stem_word, '\t', str(termid), '\n'])
f_termids.write(termid_str)
termid += 1
stem_list.append(stem_word)
for token in doc_words:
doc_idx_str = ''.join([str(docid), '\t', str(all_tokens[token])])
position = 1
for s in stem_list:
if token == s:
doc_idx_str = ''.join([doc_idx_str.strip(), '\t', str(position)])
