def get_stopwords(filename):
stopwords = []
f = open(filename, 'r')
for line in f:
line = line.strip(" \n\t")
stopwords.append(line)
f.close()
return stopwords
def getTermsAndStopList(self, str):
stopwords = []
words = []
tokens = self.tokenizer.tokenize(str)
for token in tokens:
if token.isdigit() == False:
if token not in self.stop:
words.append(token)
else:
stopwords.append(token)
terms = [self.stemmer.stem(word) for word in words]
return terms, stopwords
def read(self, list, path):
#read the file (forcing utf8) and output a list
lemmer = WordNetLemmatizer()
#count the words ,save those that have a frequency of 15 or less
#so that the algorithm doesn't get distracted by irrellevant non-recurrent words
data_file = open(path, 'rt', encoding='utf8', errors='replace')
cn = Counter(word for l in data_file for word in l.split())
words = dict((word, v) for word, v in cn.items() if v < 6)
words_list = words.keys()
data_file.close()
#read the stopwords file and create a list with them
stopwords_f = open('stopwords.txt',
'rt',
encoding='utf8',
errors='replace')
stopwords = []
for line in stopwords_f:
stopwords.append(str(line.strip()))
stopwords_f.close()
data_file = open(path, 'rt', encoding='utf8', errors='replace')
for line in data_file:
#manually check for stopwords
new_line = ''
#manually eliminate puntuation and grammar stuff, according to stastics
#of the data, least repeated and most repeated words (irrellevant ones)
line = str(line).replace(",", " ")
line = str(line).replace(' " ', ' ')
line = str(line).replace(".", "")
line = str(line).replace(" ?", "")
line = str(line).replace(" : ", " ")
line = str(line).replace(" ; ", " ")
line = str(line).replace(" ( ", " ")
line = str(line).replace(" ) ", " ")
line = str(line).replace(". . .", " ")
line = str(line).replace(" -- ", " ")
#remove some stopwords manually as well as digits from the strings
line = self.replaceMultiple(str(line), words_list, ' ')
#remove least used words (words that have a frequency of 3 and less)
line = self.replaceMultiple(str(line), stopwords, ' ')
#lematize the final words in the line
for word in line.split(' '):
new_line = new_line + ' ' + str(lemmer.lemmatize(word))
list.append(line.rstrip())
data_file.close()
#output the lsit
return list
def remove_stop_word(tweet):
stopwords = []
with open("english") as files:
for line in files:
values = line.split()
word = values[0]
stopwords.append(word)
pattern = re.compile(r'\b(' + r'|'.join(stopwords) + r')\b\s*')
tweet = pattern.sub('', tweet)
return tweet
def remove_stopwords(df):
""" Removes stopwords based on a known set of stopwords
available in the nltk package. In addition, we include our
made up word in here.
"""
# Luckily nltk already has a set of stopwords that we can remove from the texts.
stopwords = nltk.corpus.stopwords.words('english')
# we'll add our own special word in here 'qwerty'
stopwords.append(our_special_word)
df['stopwords_removed'] = list(map(lambda doc:
[word for word in doc if word not in stopwords],
df['tokenized_text']))
def get_stopwords():
initial_stopwords = [
"a", "about", "above", "across", "after", "afterwards", "again",
"against", "all", "almost", "alone", "along", "already", "also",
"although", "always", "am", "among", "amongst", "amoungst", "amount",
"an", "and", "another", "any", "anyhow", "anyone", "anything",
"anyway", "anywhere", "are", "around", "as", "at", "back", "be",
"became", "because", "become", "becomes", "becoming", "been", "before",
"beforehand", "behind", "being", "below", "beside", "besides",
"between", "beyond", "bill", "both", "bottom", "but", "by", "call",
"can", "cannot", "cant", "co", "con", "could", "couldnt", "cry", "de",
"describe", "detail", "did", "do", "does", "done", "down", "due",
"during", "each", "eg", "eight", "either", "eleven", "else",
"elsewhere", "empty", "enough", "etc", "even", "ever", "every",
"everyone", "everything", "everywhere", "except", "few", "fifteen",
"fifty", "fill", "find", "fire", "first", "five", "for", "former",
"formerly", "forty", "found", "four", "from", "front", "full",
"further", "get", "give", "go", "had", "has", "hasnt", "have", "he",
"hence", "her", "here", "hereafter", "hereby", "herein", "hereupon",
"hers", "herself", "him", "himself", "his", "how", "however",
"hundred", "i", "ie", "if", "in", "inc", "indeed", "interest", "into",
"is", "it", "its", "itself", "keep", "last", "latter", "latterly",
"least", "less", "ltd", "made", "many", "may", "me", "meanwhile",
"might", "mill", "mine", "more", "moreover", "most", "mostly", "move",
"much", "must", "my", "myself", "name", "namely", "neither", "never",
"nevertheless", "next", "nine", "no", "nobody", "none", "noone", "nor",
"not", "nothing", "now", "nowhere", "of", "off", "often", "on", "once",
"one", "only", "onto", "or", "other", "others", "otherwise", "our",
"ours", "ourselves", "out", "over", "own", "part", "per", "perhaps",
"please", "put", "rather", "re", "same", "see", "seem", "seemed",
"seeming", "seems", "serious", "several", "she", "should", "show",
"side", "since", "sincere", "six", "sixty", "so", "some", "somehow",
"someone", "something", "sometime", "sometimes", "somewhere", "still",
"such", "system", "take", "ten", "than", "that", "the", "their",
"them", "themselves", "then", "thence", "there", "thereafter",
"thereby", "therefore", "therein", "thereupon", "these", "they",
"thick", "thin", "third", "this", "those", "though", "three",
"through", "throughout", "thru", "thus", "to", "together", "too",
"top", "toward", "towards", "twelve", "twenty", "two", "un", "under",
"until", "up", "upon", "us", "very", "via", "was", "we", "well",
"were", "what", "whatever", "when", "whence", "whenever", "where",
"whereafter", "whereas", "whereby", "wherein", "whereupon", "wherever",
"whether", "which", "while", "whither", "who", "whoever", "whole",
"whom", "whose", "why", "will", "with", "within", "without", "would",
"yet", "you", "your", "yours", "yourself", "yourselves"
]
stopwords = list()
for s in initial_stopwords:
s = normalize_string(s)
stopwords.append(s)
return initial_stopwords
def load_en_stopwords(filename):
'''Loads English stop-words from a given file
Return:
a list of stop words
Arguments:
the stop-words file name
'''
stopwords = []
with codecs.open(filename, mode='r', encoding='utf-8') as fSW:
for line in fSW:
stopwords.append(line.strip().lower())
return stopwords
def ques23(entiresentence,string,np,x,y):
stopwords=[]
querywords = string.split()
#print("querywords",querywords)
grammar = "chunk:{<VB.?|MD|RP|RB.?>+<DT>?<RB.?>*<JJ.?>*<NN.?|PRP|PRP$|POS|VBG|DT|CD|VBN>+}"
res,sentence=findChunk(string,grammar,"chunk")
#print("res",res,"sentence",sentence)
# for subtree in sentence.subtrees():
# if subtree.label() == 'chunk':
# print(subtree)
# for pr in np:
# if()
if(len(sentence)>0):
for words in sentence:
stopwords.append(words[0])
#print("sp",stopwords)
stopwords.append(np)
stopwords.append(x)
stopwords.append(y)
stopwords=set(stopwords)
#print(stopwords)
resultwords = [word for word in querywords if word not in stopwords]
resultwords = ' '.join(str(e) for e in resultwords)
#print("rs",resultwords)
WhatQues="What "+x+" "+np+" "+y+" "+resultwords+"?"
#print(WhatQues)
AnsQues[entiresentence].append(WhatQues)
def get_hindi_stopwords(filename="stop-words.txt"):
"""Get stopwords in Hindi.
Args:
filename - stopwords_set words file
Returns:
set of hindi stop words
"""
stopwords = []
with open(filename, "r") as stop_words:
for word in stop_words.readlines():
stopwords.append(unidecode(word.strip().decode("utf8")))
return set(stopwords)
def preprocessing(tagged_by_Sentence):
"""
1. 특수문자 제거, 소문자
2. not, "n't" -> not_stemming(다음단어)
3. 특수문자 제거, 숫자 제거
4. stopword 제거
5. stemming
"""
from nltk.corpus import stopwords
stopwords = stopwords.words('english')
stopwords.remove("not")
stopwords.remove('very')
stopwords.append("'m")
stopwords.append("'s")
re_special = re.compile('[^A-Za-z0-9]+') # 문자,숫자 제외한 나머지
re_num = re.compile('[0-9]+') # 숫자
st = PorterStemmer()
new_sent = []
not_indice = []
for sent in tagged_by_Sentence:
text = [(tup[0].lower(), tup[1]) for tup in sent
if not bool(re_special.match(tup[0]))] # 1. 특수문자 제거, 소문자
# 2. not, n't 랑 다음단어 합치기
# not, n't 가 나오면 다음 단어랑 합치고, 그 다음 단어의 index를 저장해놨다가 del_element_by_indice 함수에서 제거
new_text = []
for index, tup in enumerate(text):
if tup[0] == "n't" or tup[0] == "not":
if index + 1 < len(text):
if not bool(re_special.match(
text[index + 1][0])) or text[index + 1][1] != 'CD':
new_text.append("not_" + st.stem(text[index + 1][0]))
not_indice.append(index)
else:
new_text.append("not")
else:
if not bool(re_num.match(
tup[0])) or tup[1] != 'CD': # 3. 특수문자, 숫자 제거
new_text.append(tup[0])
new_text = del_element_by_indice(new_text, not_indice)
new_words = [
st.stem(word) for word in new_text if word not in stopwords
] # 4,5 stopword 제거, stemming
new_sent.append(new_words)
return new_sent
def tokenize_headlines_with_sentiment(df):
headlines = df.title.tolist()
all_bigrams = []
headlines_string = (' '.join(filter(None, headlines))).lower()
tokens = word_tokenize(headlines_string)
# Remove single letter tokens
tokens_sans_singles = [i for i in tokens if len(i) > 1]
# Remove stop words
stopwords = nltk.corpus.stopwords.words('english')
new_words=("s'","'s","election", "2020", "n't", "wo","...", "'")
for i in new_words:
stopwords.append(i)
tokens_sans_stop = [t for t in tokens_sans_singles if t not in stopwords]
tokens_sans_stop = [t.replace('wins', 'win') for t in tokens_sans_stop]
# Get bigrams and frequencies
bi_grams = list(ngrams(tokens_sans_stop, 2))
counter = Counter(bi_grams)
# Convert counter dictionary to dataframe
counter_df = pd.DataFrame.from_dict(counter, orient='index').reset_index().rename(columns={"index": "bigram", 0: "freq"})
counter_df_sort = counter_df.sort_values(by=['freq'],ignore_index=True, ascending=False)
# Create concatenated bigram string for sentiment scoring
counter_df_sort['word1'], counter_df_sort['word2'] = counter_df_sort.bigram.str
counter_df_sort['bigram_joined'] = counter_df_sort.word1 + " " + counter_df_sort.word2
counter_df_sort=counter_df_sort.drop(['word1','word2'], axis=1)
# get sentiment for bigrams
analyzer = SentimentIntensityAnalyzer()
bigrams_scores = counter_df_sort['bigram_joined'].apply(analyzer.polarity_scores).tolist()
df_bigrams_scores = pd.DataFrame(bigrams_scores).drop(['neg','neu','pos'], axis=1).rename(columns={"compound": "sentiment_compound"})
bigrams_freq_and_scores = counter_df_sort.join(df_bigrams_scores, rsuffix='_right')
print(f"There are {len(bigrams_freq_and_scores)} extracted bigrams across all headlines")
return bigrams_freq_and_scores
def stem_tokenize(str_use):
"""
Takes a string and tokenizes it, stripping it of punctuation and stopwords. Returns a list of strings.
"""
stopwords = nltk.corpus.stopwords.words('english')
stopwords.extend(string.punctuation)
stopwords.append('')
addstopwords = ["in", "on", "of", "''"]
stopwords.append(addstopwords)
stemmer = wordnet.WordNetLemmatizer()
tokenizer = punkt.PunktWordTokenizer()
# removes stopwords and punctuation, then splits the string into a list of words
token = [token.lower().strip(string.punctuation) for token in tokenizer.tokenize(str_use)
if token.lower().strip(string.punctuation) not in stopwords]
text = [word for word in token if re.search(r'[a-zA-Z]', word) is not None]
stem = [stemmer.lemmatize(word) for word in text]
# Returns a list of strings
return stem
def count_words(text):
stopwords = nltk.corpus.stopwords.words('english')
stopwords.append('https')
stopwords.append('http')
stopwords.append('im')
stopwords.append('# ')
# RegEx for stopwords
RE_stopwords = r'\b(?:{})\b'.format('|'.join(stopwords))
# replace '|'-->' ' and drop all stopwords
return (text.str.lower().replace([r'\|', RE_stopwords], [' ', ''],
regex=True).str.cat(sep=' ').split())
def displaywordcloud(data=None,
backgroundcolor='#fff',
width=1000,
height=1000):
stopwords.append(product_name)
wordcloud = WordCloud(
font_path='C:Windows/Fonts/NanumGothicCoding-Bold.ttf',
mask=mask_png,
stopwords=stopwords,
collocations=False,
max_font_size=160,
colormap='tab10',
background_color=backgroundcolor,
width=width,
height=height).generate(data)
fig = plt.figure(figsize=(10, 10))
plt.imshow(wordcloud, interpolation="bilinear", aspect='auto')
plt.axis("off")
# plt.show
fig.savefig('webservice/static/wordcloud.png')
def Cleaning(liste):
import re
import nltk
nltk.download('stopwords')
from nltk.corpus import stopwords
from nltk.stem.porter import PorterStemmer
ps = PorterStemmer()
stopwords = list(set(stopwords.words('english')))
Liste = ['a','b','c','d','e','f','g','h','i','j','k','l','m','n','o','p','q','r','s','t','u','v','w','x','y','z']
for i in range(0,len(Liste)):
stopwords.append(Liste[i])
corpus = []
for i in range(0,len(liste),1):
review = re.sub('[^a-zA-Z]',' ',liste[i])
review = review.lower()
review = review.split()
review = [ps.stem(i) for i in review if i not in stopwords]
review = ' '.join(review)
corpus.append(review)
return corpus
def is_ci_stem_stopword_set_match(self, a, b, threshold=0.5):
# Get default English stopwords and extend with punctuation
stopwords = nltk.corpus.stopwords.words('english')
stopwords.extend(string.punctuation)
stopwords.append('')
# Create tokenizer and stemmer
tokenizer = nltk.tokenize.punkt.PunktWordTokenizer()
stemmer = nltk.stem.snowball.SnowballStemmer('english')
"""Check if a and b are matches."""
tokens_a = [token.lower().strip(string.punctuation) for token in tokenizer.tokenize(a) \
if token.lower().strip(string.punctuation) not in stopwords]
tokens_b = [token.lower().strip(string.punctuation) for token in tokenizer.tokenize(b) \
if token.lower().strip(string.punctuation) not in stopwords]
stems_a = [stemmer.stem(token) for token in tokens_a]
stems_b = [stemmer.stem(token) for token in tokens_b]
# Calculate Jaccard similarity
ratio = len(set(stems_a).intersection(stems_b)) / float(
len(set(stems_a).union(stems_b)))
return (ratio >= threshold)
def nounphrase(tree):
stopwords=[]
querywords=[]
grammar = "verb:{<VBG|VBN|VB.?|MD|RP>+}"
res,verbtree=findChunkwithPOSTags(cstr,grammar,"verb")
# print("res",res,"sentence",verbtree)
#print(len(verbtree))
if(len(verbtree)>0):
for subtree in verbtree.subtrees():
for words in subtree:
stopwords.append(words[0])
for subtree in tree.subtrees():
for words in subtree:
querywords.append(words[0])
# print("querywords",querywords)
resultwords = [word for word in querywords if word not in stopwords]
resultwords = ' '.join(str(e) for e in resultwords)
#print("rs",str(resultwords))
return 1,resultwords
else:
return 0,""
def update_stopwords(self,
add_words=[],
remove_words=[],
update_corpus=True):
stopwords = self.stopwords
[stopwords.append(x) for x in add_words]
[stopwords.remove(x) for x in remove_words if x in stopwords]
self._stopwords_ = stopwords
if update_corpus:
self.prepare_corpus()
def ques2_2(entiresentence,string,np,x,y):
np=""
stopwords=[]
querywords = string.split()
grammar = "chunk:{<IN>+<DT>?<RB.?>*<JJ.?>*<NN.?|PRP|PRP$|POS|VBG|DT|CD|VBN>+}"
res,sentence=findChunk(string,grammar,"chunk")
if(res!=0):
for words in sentence:
stopwords.append(words[0])
#print("sp",stopwords)
stopwords.append(np)
stopwords.append(x)
# stopwords.append(y)
#print(stopwords)
########FIND THE PREPOSTION
preposition=""
for words in sentence:
#print("words in sentence",words)
if(words[1]=='IN'):
#print("prep found",words[0])
preposition=words[0]
#print("preposition is",preposition)
#print("///////////////////////////////////")
resultwords = [word for word in querywords if word not in stopwords]
resultwords = ' '.join(str(e) for e in resultwords)
#print(resultwords)
prepques=preposition+" what "+x+np+" "+resultwords+"?"
#print(prepques)
AnsQues[entiresentence].append(prepques)
def getStopWordList(stopWordFile):
stopwords = []
stopwords.append("AT_USER")
stopwords.append("URL")
with open(stopWordFile, 'r') as f:
reader = csv.reader(f)
for w in reader:
stopwords.append(w[0])
return stopwords
print("Dataset Loaded...\n")
# Tokenization and Stemming
stemmer = SnowballStemmer('english')
tokenizer = RegexpTokenizer(r'[a-zA-Z\']+')
def tokenize(text):
return [stemmer.stem(word) for word in tokenizer.tokenize(text.lower())]
#Adding new stopwords
stopwords = nltk.corpus.stopwords.words('english')
with open('StopWords.txt', 'r') as file:
for i in file:
stopwords.append(i.strip())
# Feature extraction using TF-IDF
tfidf = TfidfVectorizer(stop_words=stopwords, tokenizer=tokenize)
X = tfidf.fit_transform(data)
words = tfidf.get_feature_names()
print(words)
print("Number of features: " + str(len(words))) #check number of features
print("Vectorization Completed...\n")
'''
# Clustering using K-Means
print("k-Means with 5 cluster:\n")
kmeans_1 = KMeans(init='k-means++', n_clusters = 5, random_state = 42)
kmeans_1.fit(X)
common_words = kmeans_1.cluster_centers_.argsort()[:,-1:-21:-1]
for num, centroid in enumerate(common_words):
import requests
import string
#reads from the file defined in project specs on ecampus
url = "http://www.gutenberg.org/cache/epub/9845/pg9845.txt"
r = requests.get(url, allow_redirects=True)
# alternate:
# raw = "your string here" or read in a txt file
#tokenize to keep only words
tokenizer = RegexpTokenizer(r'\w+')
tokens = tokenizer.tokenize(r.text)
#get stopwords + extra for the era of the book at the url
stopwords = stopwords.words('english')
extrastop = ["thee", "thy", "ye", "thine", "thou"]
stopwords.append(extrastop)
nostop = [w for w in tokens if not w in stopwords]
#remove repeats
norepeat = list(set(nostop))
#add all of the words + their initial location to a dictionary
dictionary = {}
for w in norepeat:
dictionary[w.upper().encode('utf8')] = r.text.find(w)
#write to file
with open('tokens.txt', 'w') as file:
file.write(str(dictionary))
def FeaturizeFile(df):
# df = pd.read_csv(CSVfile,encoding = 'latin1')
stats = []
attribute_name = []
sample = []
id_value = []
i = 0
castability = []
number_extraction = []
avg_tokens = []
ratio_dist_val = []
ratio_nans = []
keys = list(df.keys())
attribute_name.extend(keys)
summary_stat_result = summary_stats(df, keys)
stats.extend(summary_stat_result)
samples = get_sample(df, keys)
sample.extend(samples)
# castability.extend(castability_feature(df, keys))
# number_extraction.extend(numeric_extraction(df, keys))
# avg_tokens.extend(get_avg_tokens(samples))
ratio_dist_val.extend(get_ratio_dist_val(summary_stat_result))
ratio_nans.extend(get_ratio_nans(summary_stat_result))
csv_names = [
'Attribute_name', 'total_vals', 'num_nans', 'num_of_dist_val', 'mean',
'std_dev', 'min_val', 'max_val', '%_dist_val', '%_nans', 'sample_1',
'sample_2', 'sample_3', 'sample_4', 'sample_5'
]
golden_data = pd.DataFrame(columns=csv_names)
for i in range(len(attribute_name)):
# print(attribute_name[i])
val_append = []
val_append.append(attribute_name[i])
val_append.extend(stats[i])
val_append.append(ratio_dist_val[i])
val_append.append(ratio_nans[i])
val_append.extend(sample[i])
# val_append.append(castability[i])
# val_append.append(number_extraction[i])
# val_append.append(avg_tokens[i])
golden_data.loc[i] = val_append
# print(golden_data)
curdf = golden_data
for row in curdf.itertuples():
# print(row[11])
is_list = False
curlst = [row[11], row[12], row[13], row[14], row[15]]
delim_cnt, url_cnt, email_cnt, date_cnt = 0, 0, 0, 0
chars_totals,word_totals,stopwords,whitespaces,delims_count = [],[],[],[],[]
for value in curlst:
word_totals.append(len(str(value).split(' ')))
chars_totals.append(len(str(value)))
whitespaces.append(str(value).count(' '))
if del_reg.match(str(value)): delim_cnt += 1
if url_reg.match(str(value)): url_cnt += 1
if email_reg.match(str(value)): email_cnt += 1
delims_count.append(len(delimeters.findall(str(value))))
tokenized = word_tokenize(str(value))
# print(tokenized)
stopwords.append(len([w for w in tokenized if w in stop_words]))
try:
_ = pd.Timestamp(value)
date_cnt += 1
except ValueError:
date_cnt += 0
# print(delim_cnt,url_cnt,email_cnt)
if delim_cnt > 2: curdf.at[row.Index, 'has_delimiters'] = True
else: curdf.at[row.Index, 'has_delimiters'] = False
if url_cnt > 2: curdf.at[row.Index, 'has_url'] = True
else: curdf.at[row.Index, 'has_url'] = False
if email_cnt > 2: curdf.at[row.Index, 'has_email'] = True
else: curdf.at[row.Index, 'has_email'] = False
if date_cnt > 2: curdf.at[row.Index, 'has_date'] = True
else: curdf.at[row.Index, 'has_date'] = False
curdf.at[row.Index, 'mean_word_count'] = np.mean(word_totals)
curdf.at[row.Index, 'std_dev_word_count'] = np.std(word_totals)
curdf.at[row.Index, 'mean_stopword_total'] = np.mean(stopwords)
curdf.at[row.Index, 'stdev_stopword_total'] = np.std(stopwords)
curdf.at[row.Index, 'mean_char_count'] = np.mean(chars_totals)
curdf.at[row.Index, 'stdev_char_count'] = np.std(chars_totals)
curdf.at[row.Index, 'mean_whitespace_count'] = np.mean(whitespaces)
curdf.at[row.Index, 'stdev_whitespace_count'] = np.std(whitespaces)
curdf.at[row.Index, 'mean_delim_count'] = np.mean(whitespaces)
curdf.at[row.Index, 'stdev_delim_count'] = np.std(whitespaces)
if curdf.at[row.Index,
'has_delimiters'] and curdf.at[row.Index,
'mean_char_count'] < 100:
curdf.at[row.Index, 'is_list'] = True
else:
curdf.at[row.Index, 'is_list'] = False
if curdf.at[row.Index, 'mean_word_count'] > 10:
curdf.at[row.Index, 'is_long_sentence'] = True
else:
curdf.at[row.Index, 'is_long_sentence'] = False
# print(np.mean(stopwords))
# print('\n\n\n')
golden_data = curdf
return golden_data
for word in text:
if word in stopwords_set:
continue
if words.get(word) is None:
words[word] = 1
else:
words[word] = words[word] + 1
word_lis = []
for word, no in words.items():
word_lis.append([word, no])
word_lis = pd.DataFrame(word_lis, columns=["word", "freq"])
word_lis = word_lis.sort_values(by="freq", ascending=False)
for word in word_lis['word'][:20]:
stopwords.append(word)
# print stopwords
# In[5]:
## Remove stop words
def s_wor_rm(text):
words = []
text = text.split()
for word in text:
if word in stopwords:
continue
words.append(word)
def http (text): stopwords=[] stopwords.append([w for w in text if 'https' and 'http' in w ]) mynewtext = [w for w in text if w not in stopwords[0]] return mynewtext
def remove_hashtag(sentence):
stopwords=[]
base = tokinizer(sentence)
stopwords.append([w for w in base if w.startswith('#')])
mynewtext = [w for w in base if w not in stopwords[0]]
return WordList_to_sentence(mynewtext)
dict[subject] = count
return dict
if __name__ == "__main__":
tokenizer = RegexpTokenizer(r'\w+')
stemmer = SnowballStemmer("english")
## get stop words and normalize
initial_stopwords = stopwords()
stopwords = list()
for s in initial_stopwords:
s = normalize_string(s)
stopwords.append(s)
print("Loading index")
mention_dict = load_index("../data/surface_forms_new.txt")
subject_predicates_dict = [
] # load_subject_predicates("data/SimpleQuestions_v2/freebase-FB2M.txt")
subject_triple_counts = [
] #load_subject_triple_counts("data/subject_triple_counts.txt")
dataset_names = ["test"]
max_ngram_size = 10
exclude_small_ngrams = True
exclude_stop_words = True
for d in dataset_names:
correct_count = 0
from ML_models import train_svm,train_knn,knn_accuracies,train_log_regression,\
log_regression_stats,svm_stats,train_random_forest
from dataset_builder import add_sub_pol_to_dataset
import matplotlib.pyplot as plt
import plotly.express as px
import seaborn as sns
from wordcloud import WordCloud
from nltk.corpus import stopwords
###################### Settings ######################
root_path = os.path.dirname(os.path.realpath(__file__))
ds_path = os.path.join(root_path, '..', 'common', 'dataset.csv')
clean_ds_path = os.path.join(root_path, '..', 'common', 'clean_dataset.csv')
stopwords = stopwords.words('french')
stopwords.append('a')
stopwords.append('e')
stopwords.append('Tre')
stopwords.append('cest')
amazon_img_path = 'https://i1.wp.com/www.joptimisemonsite.fr/wp-content/uploads/2015/02/logo-amazon.jpg?fit=810%2C295&ssl=1&is-pending-load=1'
#######################################################
###################### Utils ######################
@st.cache
def load_data():
ds = pd.read_csv(ds_path)
clean_ds = pd.read_csv(clean_ds_path)
clean_ds = add_sub_pol_to_dataset(clean_ds)
clean_ds = clean_ds.drop(clean_ds[clean_ds.Subjectivity > 1.0].index)
return ds, clean_ds
def cluster_topics(json_data):
import json
import time
import nltk
import re
import numpy as np
import scipy as sp
import math
from nltk.corpus import stopwords
from nltk.stem.snowball import SnowballStemmer
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity
from sklearn.cluster import KMeans
from sklearn.preprocessing import normalize
def tokenize_and_stem(text):
# first tokenize by sentence, then by word to ensure that punctuation is caught as it's own token
tokens = [word for sent in nltk.sent_tokenize(text) for word in nltk.word_tokenize(sent)]
filtered_tokens = []
# filter out any tokens not containing letters (e.g., numeric tokens, raw punctuation)
for token in tokens:
if re.search('[a-zA-Z]', token) and token not in stopwords:
filtered_tokens.append(token)
stems = [stemmer.stem(t) for t in filtered_tokens]
return stems
def tokenize_only(text):
# first tokenize by sentence, then by word to ensure that punctuation is caught as it's own token
tokens = [word.lower() for sent in nltk.sent_tokenize(text) for word in nltk.word_tokenize(sent)]
filtered_tokens = []
# filter out any tokens not containing letters (e.g., numeric tokens, raw punctuation)
for token in tokens :
if re.search("[a-zA-Z]", token) and token not in stopwords:
filtered_tokens.append(token)
return filtered_tokens
#data = json.loads(open("room_messages.json").read())
data = json.loads(open(json_data).read())
stopwords = stopwords.words("english")
stopwords.append('blah')
stemmer = SnowballStemmer("english")
# Extract messages
messages = []
time_stamps = []
for i in range(len(data["items"])):
if "text" in data["items"][i]:
message = (data["items"][i]["text"])
time_stamp = data["items"][i]["created"]
t = float(re.search("[0-9]*:[0-9]*", time_stamp).group(0).replace(":", "."))
messages.append(message)
time_stamps.append(t)
# Extract stemmed and tokenized vocab
totalvocab_stemmed = []
totalvocab_tokenized = []
for i in messages:
allwords_stemmed = tokenize_and_stem(i) # for each item in 'synopses', tokenize/stem
totalvocab_stemmed.extend(allwords_stemmed) # extend the 'totalvocab_stemmed' list
allwords_tokenized = tokenize_only(i)
totalvocab_tokenized.extend(allwords_tokenized)
# vocab_frame = pd.DataFrame({'words': totalvocab_tokenized}, index=totalvocab_stemmed)
# print('there are ' + str(vocab_frame.shape[0]) + ' items in vocab_frame')
# Cluster messages according to topic using K-means
# define vectorizer parameters
if len(allwords_stemmed) > 200:
tfidf_vectorizer = TfidfVectorizer(max_df=0.99, max_features=200000,
min_df=0.01, stop_words='english',
use_idf=True, tokenizer=tokenize_and_stem, ngram_range=(1, 3))
else:
tfidf_vectorizer = TfidfVectorizer(max_df=1.0, max_features=200000,
min_df=0.0, stop_words='english',
use_idf=True, tokenizer=tokenize_and_stem, ngram_range=(1, 3))
# fit the vectorizer to messages
# t1 = time.clock()
tfidf_matrix = tfidf_vectorizer.fit_transform(messages)
# t2 = time.clock()
# print("Tf-Idf fit time: ", t2-t1)
# print(tfidf_matrix.shape)
terms = tfidf_vectorizer.get_feature_names()
# insert additional time feature
new_column = np.asarray(time_stamps).reshape(-1, 1)
time_norm = normalize(new_column)
# print(new_column.shape)
final = sp.sparse.hstack((tfidf_matrix, new_column))
# print(final.shape)
terms.append('time')
# calculate distance between messages using cosine similarity of tf-idf
dist = 1 - cosine_similarity(final)
# K-means clustering
# num_clusters = 5
if len(messages) > 10:
num_clusters = math.floor(math.sqrt(len(messages)) / 2)
else:
num_clusters = 1
km = KMeans(n_clusters=num_clusters)
t3 = time.clock()
km.fit(final)
t4 = time.clock()
#print("K-means fit time: ", t4 - t3)
clusters = km.labels_.tolist()
topics = {}
for t in range(num_clusters):
t_name = "topic" + str(t)
topics[t_name] = {}
t_messages = []
for i in range(len(messages)):
if clusters[i] == t:
t_messages.append(messages[i])
topics[t_name]['messages'] = t_messages
#Export as json
# with open('topics.json', 'w') as outfile:
#json.dump(topics, outfile)
# Inspect clusters
# sorted_messages = {'message': messages, 'cluster': clusters}
#
# frame = pd.DataFrame(sorted_messages, index=[clusters], columns=['message', 'cluster'])
# print(frame['cluster'].value_counts())
#
# # top words per cluster
# print("Top terms per cluster:")
# print()
# # sort cluster centers by proximity to centroid
# order_centroids = km.cluster_centers_.argsort()[:, ::-1]
#
# for i in range(num_clusters):
# print("Cluster %d words:" % i, end='')
#
# for ind in order_centroids[i, :6]: # replace 6 with n words per cluster
# print(' %s' % vocab_frame.ix[terms[ind].split(' ')].values.tolist()[0][0].encode('utf-8', 'ignore'), end=',')
# print() # add whitespace
# print() # add whitespace
#
# print("Cluster %d messages:" % i, end='')
# for message in frame.ix[i]['message'].values.tolist():
# print(' %s,' % message, end='')
# print() # add whitespace
# print() # add whitespace
return topics
from nltk.corpus import stopwords
from nltk.tokenize import wordpunct_tokenize
from sklearn.feature_extraction.text import CountVectorizer
from scipy.sparse.csr import csr_matrix
import re, pdb, os, json
import numpy as np
from collections import Counter
import nltk
np.random.seed(1234)
lemmatizer = WordNetLemmatizer()
pattern_train = r"(\d*),\"(.*)\",(\d*)"
pattern_test = r"(\d*),\"(.*)\""
stopwords = list(stopwords.words("english"))
stopwords.append("__EOS__")
sw = [
"right",
"love",
"people",
"feel",
"yeah",
"one",
"see",
"something",
"want",
"year",
"yes",
"still",
"kind",
from nltk import bigrams, trigrams
import math
import json
from nltk.stem import WordNetLemmatizer
stopwords = nltk.corpus.stopwords.words('english')
tokenizer = RegexpTokenizer("[\w’]+", flags=re.UNICODE)
#st = LancasterStemmer()
wnl = WordNetLemmatizer()
keywords=[]
with open('keywords.txt','r') as f:
for i in f:
keywords.append(i.strip())
with open('stopwords.txt','r') as f:
for i in f:
stopwords.append(i.strip())
def freq(word, doc):
return doc.count(word)
def word_count(doc):
return len(doc)
def tf(word, doc):
return (freq(word, doc) / float(word_count(doc)))
def calcu_tf(keyword):
url = "http://en.wikipedia.com/wiki/"+keyword
content = urllib2.urlopen(url).read()
def get_ngram(filename = None,_type=None,is_stopword=None):
file_content = open(filename).read()
# Get the tockens. Use word punctuations for tokenizing too other than spaces
tokens = nltk.word_tokenize(file_content)
text = nltk.Text(tokens)
word_filter = lambda *w: word_to_find not in w
## Bigrams
# ENABLE STOP WORDS
#stopwords = stopwords.words('english')
stopwords = []
stopwords.append('.')
#stopwords.append('The')
stopwords.append(':')
stopwords.append(',')
stopwords.append(';')
stopwords.append('`')
stopwords.append('``')
stopwords.append('\"')
#print stopwords
if is_stopword == 1:
filtered_tokens = []
for ftoken in tokens:
ftoken_low = ftoken.lower()
if ftoken not in stopwords:
#print "Removing ######### " + ftoken
filtered_tokens.append(ftoken)
else:
filtered_tokens = tokens
if _type == 2:
finder = BigramCollocationFinder.from_words(filtered_tokens,window_size=3)
else:
finder = TrigramCollocationFinder.from_words(filtered_tokens,window_size=3)
# only bigrams that appear 3+ times
finder.apply_freq_filter(1)
lst = list(finder.ngram_fd.viewitems())
#print lst
ll = sorted(lst, key=lambda x: x[1])
#ll = lst.sort(key=lambda x: x[0])
#print ll
res =[]
for i in ll:
k1 = []
k = list(i)
k1.append(' '.join(k[0]))
k1.append(k[1])
res.append(k1)
#raw_input()
#print k1
#print res
return res
rows = cursor.fetchall()
except:
print "Pid : ",pid," not found"
return
return rows
#print nltk.pos_tag(['flipkart','samsung'])
#lemmatiser and stopwords initialization
lemmatizer = nltk.WordNetLemmatizer()
from nltk.corpus import stopwords
#building the stopwords list
stopwords = stopwords.words('english')
stoplist = ['>','<','%','.','br/','(',')','=','!']
for i in stoplist:
stopwords.append(i)
#normalise each qualified word
def normalise(word):
"""Normalises words to lowercase and stems and lemmatizes it."""
word = word.lower()
word = lemmatizer.lemmatize(word)
return word
def armAssoc(dust,strength):
print
def writetofile(l):
feat=[]
remove=["flipkart","problem","time","product","awesome","thing","port","delivery","buying","perfect","mode","reason","anything","point","excellent","hand","till","fact","market","weather","brand","life","option","guide","money"]
for i in range(0,len(l)):
import bs4 as bs
import json
import re
import nltk
import heapq
import pickle
import sys
from pprint import pprint
import os
import codecs
import string
from sklearn_crfsuite import metrics
from DataExtraction import convertCONLLFormJustExtractionSemEvalPerfile
from FeatureExtraction import sent2labels, sent2features
from PhraseEval import phrasesFromTestSenJustExtractionWithIndex
from nltk import word_tokenize, pos_tag, ne_chunk
from nltk import conlltags2tree, tree2conlltags
#Swapped hardcoded link with system arguement
file_inLoc = sys.argv[1]
file_outLoc = sys.argv[1].split(".")[0] + "-DKE.txt"
file_outLoc = file_outLoc[15:]
with open(file_inLoc, 'r', encoding='utf-8-sig') as f:
article_text = json.load(f)
pprint(article_text)
'''
scraped_data = urllib.request.Request(file_inLoc, headers={'User-Agent' : "Magic Browser"})
scraped_data=urllib.request.urlopen(scraped_data)
parsed_article = bs.BeautifulSoup(article,'lxml')
paragraphs = parsed_article.find_all('p')
article_text = ""
words = tknzr.tokenize(words)
exclude = set(string.punctuation)
words2 = [word for word in words if
not word in exclude]
words_tag = dict(pos_tag(words))
words = [word.lower() for word in words2 if
not word in nltk.corpus.stopwords.words('english') and not word.isdigit()]
# print(words)
words = [lima(word, words) for word in words]
# print(words)
words = ' '.join(words)
# print(words)
return words
stopwords = stopwords.words('english')
stopwords.append('.')
# stopwords.union('sally')
# operators = set(('sally'))
# stop = set(nltk.corpus.stopwords.words('english')) + operators
# print(stopwords)
sentence_tokenizer = nltk.data.load('tokenizers/punkt/english.pickle')
word_tokenizer = nltk.tokenize.RegexpTokenizer(r'\w+')
chapters=[]
with open('user_posts_1641812829207516.csv') as File:
tfidfReader = csv.reader(File)
for row in tfidfReader:
chapters.append(clean(row[0]).encode('utf-8'))
num_chapters = len(chapters)
fvs_lexical = np.zeros((len(chapters), 3), np.float64)
fvs_punct = np.zeros((len(chapters), 3), np.float64)
i=1
def analyze(soup_object, csv_name):
#select only the headlines in each google search result
base = soup_object.select("div.g.card h3")
#declare empty list where I'm going to put all the headlines
headlines = []
#loop to get rid of all html and keep only headline text
for row in base:
clean = row.text
headlines.append(clean)
#print to verify headlines are clean
#print(headlines)
#empty list to store tokenized headlines
tokens = []
#loop to tokenize headlines by using clean_tokens method from earlier
for each in headlines:
clean = clean_tokens(each)
tokens.append(clean)
#print tokens to verify
#print('tokens =',tokens)
#create stopwords list from nltk and add fallout 76 as stopwords
stopwords = nltk.corpus.stopwords.words('english')
stopwords.append("fallout")
stopwords.append('76')
stopwords.append("'fallout")
#remove stopwords from tokens
filtered = []
for list in tokens:
x = []
for word in list:
if word not in stopwords:
x.append(word)
filtered.append(x)
#print to verify stopwords are gone
#print("filtered = ", filtered)
#declare empty list so I can put the tokens back into headlines without stopwords
combined = []
#put tokens back into headlines without stopwords
for list in filtered:
combined.append(" ".join(list))
#import sentiment analyzer
from nltk.sentiment.vader import SentimentIntensityAnalyzer as SIA
#create sia object I think and then create empty list to put end results in
sia = SIA()
results = []
#analyze sentiment of each headline
for line in combined:
pol_score = sia.polarity_scores(line)
results.append(pol_score)
#print to verify
#for i in results:
#print(i)
#write sentiment analysis to csv file
with open('resultsfinal.csv', 'a') as csv_file:
writer = csv.writer(csv_file)
for d in results:
writer.writerow(['compound', d['compound']])
import nltk
tweets = pd.read_csv('../data/primary_debates_cleaned.csv')
tweets = tweets.drop(['URL','Location','Date','Line'], axis=1)
tweets = tweets.loc[tweets['Speaker'].isin(['Bush', 'Carson', 'Chafee', 'Christie', 'Clinton', 'Cruz', 'Fiorina', 'Gilmore', 'Graham', 'Huckabee', 'Jindal', 'Kasich', "O'Malley", 'Pataki', 'Paul', 'Perry', 'Rubio', 'Sanders', 'Santorum', 'Trump', 'Walker', 'Webb'])]
tweets = tweets.loc[~tweets['Text'].isin(['(APPLAUSE)', '(ANTHEM)', '(BELL)', '(BOOING)', '(COMMERCIAL)', '(CROSSTALK)', '(LAUGHTER)', '(MOMENT.OF.SILENCE)', '(SPANISH)', '(VIDEO.END)', '(VIDEO.START)', '(inaudible)'])]
#print(tweets)
#print(tweets.Tweet[0])
democrat = tweets[tweets.Party == 'Democratic']
republican = tweets[tweets.Party == 'Republican']
stopwords = stopwords.words('english')
#add some unnecessary word to stopwords list
stopwords.append("rt")
stopwords.append("u")
stopwords.append("amp")
stopwords.append("w")
stopwords.append("th")
clean_democrat = []
for d in democrat.Text:
d = re.sub(r'https\S+', '', d)
d = re.sub("[^a-zA-Z]", " ", d)
d = d.lower()
d = nltk.word_tokenize(d)
d = [word for word in d if not word in set(stopwords)]
lemma = nltk.WordNetLemmatizer()
d = [lemma.lemmatize(word) for word in d]
d = " ".join(d)
def pullAllTheStops(self):
stopwords = []
for word in self.words:
if word not in self.stops:
stopwords.append(word)
return stopwords
import csv
import random
import nltk
from nltk import word_tokenize
from nltk.corpus import stopwords
stopwords = stopwords.words("english")
stopwords.append('.')
stopwords.append('?')
stopwords.append('\'')
stopwords.append(',')
stopwords.append('’')
stopwords.append(')')
stopwords.append('(')
stopwords.append('/')
class DataLoader:
""" DataLoader Utility """
def __init__(self, file):
self.file = file
self.documents = []
self.all_words = []
self.questions = []
self.answers = []
self.questions_and_answer = []
# doc - document words and dictionary - corpus words
def get_feature(self, doc, dictionary):
vector = {}
words = set(doc)
for w in dictionary:
def perturb(self, stringseq, method=0):
assert len(stringseq) > 0, "sentence must be more than one char!"
assert any(
c.lower() in string.ascii_lowercase for c in stringseq
), "the sentence has to contain at least one alphabet letter!"
tokens = nltk.wordpunct_tokenize(stringseq)
# print (tokens)
#if method is not specified, we randomly select one.
if method == 0:
method = randint(1, 4)
elif method < 0:
method = randint(1, 4 + method) # eliminate certain methods
# qwerty
if method == 1:
while True:
randidx = randint(0, len(tokens) - 1)
token = tokens[randidx]
#has to be at least one alpha letter
if not any(c.lower() in string.ascii_lowercase for c in token):
continue
res = self.pert_qwerty.perturb(token)
if res[0] == res[1].lower():
continue
fr, to = self.spans(tokens, stringseq, randidx)
return (stringseq[:fr] + res[0] + stringseq[to:], res[1],
(fr + res[2][0], res[2][1]), 0)
# drop
elif method == 2:
stopwords = []
for i, token in enumerate(tokens):
if token in self.stopwords:
stopwords.append(i)
if len(stopwords) > 0:
randidx = choice(stopwords)
token = tokens[randidx]
else:
randidx = randint(0, len(tokens) - 2)
token = tokens[randidx]
fr, to = self.spans(tokens, stringseq, randidx)
# remove a whitespace if drop the word
if to < len(stringseq) and stringseq[to] == ' ':
# trailing whitespace
return (stringseq[:fr] + stringseq[to + 1:], tokens[randidx],
(fr, 0), 2)
elif stringseq[fr - 1] == ' ':
# leading whitespace
return (stringseq[:fr - 1] + stringseq[to + 1:],
tokens[randidx], (fr - 1, 0), 1)
else:
return (stringseq[:fr] + stringseq[to + 1:], tokens[randidx],
(fr, 0), 0)
# delete chars
elif method == 3:
pos_tags = nltk.pos_tag(tokens)
cnt = 0
while True:
cnt += 1
if cnt > 5:
break
randidx = randint(0, len(tokens) - 2)
token = tokens[randidx]
if not any(c.lower() in string.ascii_lowercase for c in token):
continue
if len(token) < 2 or pos_tags[randidx] in ["TO", "SYM"]:
continue
res = self.pert_delete.perturb(token, pos_tags[randidx])
fr, to = self.spans(tokens, stringseq, randidx)
return (stringseq[:fr] + res[0] + stringseq[to:], res[1],
(fr + res[2][0], res[2][1]), 0)
return self.perturb(stringseq, -2)
# synonym
elif method == 4:
cnt = 0
while True:
cnt += 1
if cnt > 5:
# in case of dead loop
break
randidx = randint(0, len(tokens) - 2)
token = tokens[randidx]
# has to be a word
if not all(c.lower() in string.ascii_lowercase for c in token):
continue
res = self.pert_synonym.perturb(token)
if res[0] == res[1]:
continue
fr, to = self.spans(tokens, stringseq, randidx)
return (stringseq[:fr] + res[0] + stringseq[to:], res[1],
(fr, res[2][1]), 0)
return self.perturb(stringseq, -1)
dic.append(
{'구분': 'trigram', '순위': '1~20', '단어': unicode(trigram[0]) + unicode(trigram[1]) + unicode(trigram[2]),
'횟수': str(dic[19]['횟수']) + u"회 이상"})
csv_columns = ['구분', '순위', '단어', '횟수']
csv_file = filename.split('.')[0] + "freq.csv"
WriteDictToCSV(csv_file, csv_columns, dic)
stopwords = []
stoptext = open("stopwords.txt", "r")
for sw in stoptext.readlines():
sw = sw.decode('cp949')
sw = re.sub('\\n', '', sw)
stopwords.append(sw)
measures = collocations.BigramAssocMeasures()
measures2 = collocations.TrigramAssocMeasures()
def remove_values_from_list(the_list, val):
return [value for value in the_list if value != val]
pwd = os.getcwd()+'\\document'
for path, dirs, files in os.walk(pwd):
for file in files:
# coding=utf-8
import math
from textblob import TextBlob as tb
from nltk.corpus import stopwords
from nltk.tokenize import word_tokenize
#STOPWORDS
stopwords = stopwords.words('portuguese')
stopwords.append('pra')
def remove_stopwords (sentences):
phrases = []
for sentence in sentences:
#Gera tokens
tokens = []
words = word_tokenize(sentence)
for word in words:
if word.lower() not in stopwords:
tokens.append(word)
phrases.append(' '.join(tokens))
return phrases
def get_cosine(vec1, vec2):
size = len(vec1) - 1
numerator = sum([vec1[x] * vec2[x] for x in range(size)])
sum1 = sum([vec1[x]**2 for x in range(size)])
sum2 = sum([vec2[x]**2 for x in range(size)])
from tweepy import OAuthHandler
from tweepy.streaming import StreamListener
from nltk.corpus import stopwords
from nltk.tokenize import word_tokenize
nltk.download('punkt')
consumerKey="xxxx"
consumerSecret="xxxx"
accessToken="xxxx-xxxx"
accessSecret="xxxxx"
stopwords = []
for i in open("stopwords.txt"):
sword = i.rstrip('\n')
stopwords.append(sword)
global check
check = []
class listener(StreamListener):
x = 0
y = 0
def on_data(self, data):
all_data = json.loads(data)
tweet = all_data["text"]
tweet = re.sub(r"http\S+", "", tweet)
analysis = TextBlob(tweet)
stop_words = set(stopwords)
filtered_words = set(analysis.words.lower()) - stop_words # remove stop words from tweet
for i in set(filtered_words): # remove @ tags
