def __init__(self, stop_words): self.stemmer = porter_stemmer.PorterStemmer() self.stops = [] if(stop_words is not None): with open(stop_words, 'r') as file: for line in file: self.stops.append(line)
def get_words(text): exclude = set(string.punctuation) words = text.split() proc_words = [] for word in words: word = ''.join(ch for ch in word if ch not in exclude and ch.isalnum()) word = word.lower() #if not word in stop_words: word = porter_stemmer.PorterStemmer().stem(word, 0,len(word)-1) if len(word) > 0 and word != "\n" and word != "\r": proc_words.append(word) return proc_words
def stemmer(wordList: [str]) -> [str]:
'''
This stemmer takes a list of token and convert all tokens in the list to its stemmed form
This stemmer uses Porter Stemmer.
The source code can be found on https://tartarus.org/martin/PorterStemmer/
'''
myStemmer = ps.PorterStemmer()
stemmedWordList = []
for word in wordList:
if(word.isalpha()):# if word is an English word (contains only alphabets)
word = myStemmer.stem(word, 0, len(word)-1)
stemmedWordList.append(word)
return stemmedWordList
def stem(word_list):
stemmed_list = []
p = porter_stemmer.PorterStemmer()
if len(word_list) > 0:
for x in word_list:
output = ''
word = ''
line = x + "\n"
if line == '':
break
for c in line:
if c.isalpha():
word += c.lower()
else:
if word:
output += p.stem(word, 0, len(word) - 1)
word = ''
output += c.lower()
stemmed_list.append(output.strip("\n"))
#print(stemmed_list)
return stemmed_list
import os
import porter_stemmer
p = porter_stemmer.PorterStemmer()
positiveFileList = sorted(
list(filter(lambda s: s.endswith(".tag"), os.listdir("./POS"))))
negativeFileList = sorted(
list(filter(lambda s: s.endswith(".tag"), os.listdir("./NEG"))))
for f in positiveFileList:
infile = open("./POS/" + f, 'r')
outputString = ''
while 1:
output = ''
word = ''
line = infile.readline()
if line == '':
break
for c in line:
if c.isalpha():
word += c.lower()
else:
if word:
output += p.stem(word, 0, len(word) - 1)
word = ''
output += c.lower()
outputString += output
def run_samples():
"""
"""
stemmer = porter_stemmer.PorterStemmer()
words_1 = [
("caresses", "caress"),
("ponies", "poni"),
("ties", "ti"),
("caress", "caress"),
("cats", "cat"),
("feed", "feed"),
("agreed", "agree"),
("plastered", "plaster"),
("bled", "bled"),
("motoring", "motor"),
("sing", "sing"),
("conflated", "conflate"),
("troubled", "trouble"),
("sized", "size"),
("hopping", "hop"),
("tanned", "tan"),
("falling", "fall"),
("hissing", "hiss"),
("fizzed", "fizz"),
("failing", "fail"),
("filing", "file"),
("happy", "happi"),
("sky", "sky"),
]
words_2 = [
("relational", "relate"),
("conditional", "condition"),
("rational", "rational"),
("valenci", "valence"),
("hesitanci", "hesitance"),
("digitizer", "digitize"),
("conformabli", "conformable"),
("radicalli", "radical"),
("differentli", "different"),
("vileli", "vile"),
("analogousli", "analogous"),
("vietnamization", "vietnamize"),
("predication", "predicate"),
("operator", "operate"),
("feudalism", "feudal"),
("decisiveness", "decisive"),
("hopefulness", "hopeful"),
("callousness", "callous"),
("formaliti", "formal"),
("sensitiviti", "sensitive"),
("sensibiliti", "sensible"),
]
words_3 = [
("triplicate", "triplic"),
("formative", "form"),
("formalize", "formal"),
("electriciti", "electric"),
("electrical", "electric"),
("hopeful", "hope"),
("goodness", "good"),
]
words_4 = [
("revival", "reviv"),
("allowance", "allow"),
("inference", "infer"),
("airliner", "airlin"),
("gyroscopic", "gyroscop"),
("adjustable", "adjust"),
("defensible", "defens"),
("irritant", "irrit"),
("replacement", "replac"),
("adjustment", "adjust"),
("dependent", "depend"),
("adoption", "adopt"),
("homologou", "homolog"),
("communism", "commun"),
("activate", "activ"),
("angulariti", "angular"),
("homologous", "homolog"),
("effective", "effect"),
("bowdlerize", "bowdler"),
]
words_5 = [
("probate", "probat"),
("rate", "rate"),
("cease", "ceas"),
("controll", "control"),
("roll", "roll"),
]
for w1, w2 in words_1:
print(w1, w2, stemmer.step_1(w1))
for w1, w2 in words_2:
print(w1, w2, stemmer.step_2(w1))
for w1, w2 in words_3:
print(w1, w2, stemmer.step_3(w1))
for w1, w2 in words_4:
print(w1, w2, stemmer.step_4(w1))
for w1, w2 in words_5:
print(w1, w2, stemmer.step_5(w1))
def __init__(self): self.inverted_index = load_data() self.myStemmer = ps.PorterStemmer()
def preprocess(dir, stoplist_filename):
# Create Porter stemmer object for stemming words
p = porter_stemmer.PorterStemmer()
with open(stoplist_filename) as file:
stoplist = file.read().split()
topic_counts = dict()
for dirname, subdirnames, filenames in os.walk(dir):
#filenames = ['reut2-000.sgm'] # DON'T FORGET TO REMOVE THIS WHEN USING ALL ARTICLES!!!!!!!!!!!!!!!!!!!!!
for filename in filenames:
if ( os.path.splitext(filename)[-1].lower() == '.sgm' ): # Only pick out sgm files
with open(dir + '/' + filename) as file:
soup = BeautifulSoup(file, 'html.parser')
articles = soup.find_all('reuters')
for article in articles:
if ( (len(article.topics.contents) == 1) & (article.body is not None) ): # Only count articles with a single topic
topic_counts[article.topics.text] = topic_counts.get(article.topics.text, 0) + 1
sorted_topics = sorted(topic_counts, key=topic_counts.get, reverse=True)
popular_topics = sorted_topics[0:20] # Pick out top 20 occurring topics
filtered_articles = dict()
article_tokens = dict()
# Walk back through files and add files with appropriate topics to dictionary
for dirname, subdirnames, filenames in os.walk(dir):
#filenames = ['reut2-000.sgm'] # DON'T FORGET TO REMOVE THIS WHEN USING ALL ARTICLES!!!!!!!!!!!!!!!!!!!!!
for filename in filenames:
if ( os.path.splitext(filename)[-1].lower() == '.sgm' ):
with open(dir + '/' + filename) as file:
soup = BeautifulSoup(file, 'html.parser')
articles = soup.find_all('reuters')
for article in articles:
if (article.body is not None): # Don't consider articles with no body
if ( (len(article.topics.contents) == 1) & (article.topics.text in popular_topics) ):
text = replace_non_alphanumer(ascii_only(article.body.text).lower())
text_set = set(text.split(' '))
stemmed_text = stem_text(text_set, stoplist, p)
add_tokens(stemmed_text, article_tokens)
# Create new dictionary of frequent tokens
frequent_tokens = dict()
for word in article_tokens:
if article_tokens[word] >= 5:
frequent_tokens[word] = article_tokens[word]
# Sort tokens alphabetically and change values to be index of words
alphabetic_tokens = sorted(frequent_tokens)
i = 0
for word in alphabetic_tokens:
frequent_tokens[word] = i
i += 1
#Open files for writing
norm_freq_file = open('freq.csv', 'w')
norm_sqrt_freq_file = open('sqrtfreq.csv', 'w')
norm_log_freq_file = open('log2freq.csv', 'w')
class_file = open('reuters21578.class', 'w')
label_file = open('reuters21578.clabel', 'w')
write_label_file(alphabetic_tokens, label_file)
# Walk through files again to determine number of times each token is used in articles
for dirname, subdirnames, filenames in os.walk(dir):
#filenames = ['reut2-000.sgm'] # DON'T FORGET TO REMOVE THIS WHEN USING ALL ARTICLES!!!!!!!!!!!!!!!!!!!!!
for filename in filenames:
if ( os.path.splitext(filename)[-1].lower() == '.sgm' ): # Only pick out .sgm files
with open(dir + '/' + filename) as file:
soup = BeautifulSoup(file, 'html.parser')
articles = soup.find_all('reuters')
for article in articles:
if (article.body is not None):
if ( (len(article.topics.contents) == 1) & (article.topics.text in popular_topics) ):
text = replace_non_alphanumer(ascii_only(article.body.text).lower())
[freq, ind] = sparse_article(text, frequent_tokens, stoplist, p)
norm_freq = normalize(freq)
norm_sqrt_freq = normalize(1 + np.sqrt(freq))
norm_log_freq = normalize(1 + np.log2(freq))
write_freq_file(norm_freq, ind, article['newid'], norm_freq_file)
write_freq_file(norm_sqrt_freq, ind, article['newid'], norm_sqrt_freq_file)
write_freq_file(norm_log_freq, ind, article['newid'], norm_log_freq_file)
write_article_class(article.topics.text, article['newid'], class_file)
norm_freq_file.close()
norm_sqrt_freq_file.close()
norm_log_freq_file.close()
class_file.close()
label_file.close()
return
