aboutsummaryrefslogtreecommitdiff
path: root/pyaggr3g470r/duplicate.py
diff options
context:
space:
mode:
authorCédric Bonhomme <cedric@cedricbonhomme.org>2014-11-28 19:37:16 +0100
committerCédric Bonhomme <cedric@cedricbonhomme.org>2014-11-28 19:37:16 +0100
commit7af0dba1c342f099ce66fad2b4a2c61c95bbf437 (patch)
tree0a7794d7b8aaf5d2616c60c7bb5e181239e1c80e /pyaggr3g470r/duplicate.py
parentMisc fixes for the installation/uage of scipy/numpy/nltk. (diff)
downloadnewspipe-7af0dba1c342f099ce66fad2b4a2c61c95bbf437.tar.gz
newspipe-7af0dba1c342f099ce66fad2b4a2c61c95bbf437.tar.bz2
newspipe-7af0dba1c342f099ce66fad2b4a2c61c95bbf437.zip
Finall the method with nltk is really slow...
Diffstat (limited to 'pyaggr3g470r/duplicate.py')
-rw-r--r--pyaggr3g470r/duplicate.py44
1 files changed, 2 insertions, 42 deletions
diff --git a/pyaggr3g470r/duplicate.py b/pyaggr3g470r/duplicate.py
index 6220a3cb..23f4adc0 100644
--- a/pyaggr3g470r/duplicate.py
+++ b/pyaggr3g470r/duplicate.py
@@ -2,55 +2,15 @@
#-*- coding: utf-8 -*-
import itertools
-import nltk, string
-from sklearn.feature_extraction.text import TfidfVectorizer
-
import utils
-# tokenizers/punkt/english.pickle
-
-
-stemmer = nltk.stem.porter.PorterStemmer()
-remove_punctuation_map = dict((ord(char), None) for char in string.punctuation)
-
-def stem_tokens(tokens):
- return [stemmer.stem(item) for item in tokens]
-
-def normalize(text):
- """
- Remove punctuation, lowercase, stem
- """
- return stem_tokens(nltk.word_tokenize(text.lower().translate(remove_punctuation_map)))
-
-vectorizer = TfidfVectorizer(tokenizer=normalize, stop_words='english')
-
-def cosine_sim(article1, article2):
- try:
- tfidf = vectorizer.fit_transform([utils.clear_string(article1.content),
- utils.clear_string(article2.content)])
- except ValueError as e:
- raise e
- return ((tfidf * tfidf.T).A)[0,1]
-
def compare_documents(feed):
"""
Compare a list of documents by pair.
"""
- downloaded = nltk.download("punkt")
- if not downloaded:
- # Ubuntu packaged version still uses old URL
- dl = nltk.downloader.Downloader("https://nltk.github.com/nltk_data/")
- dl.download("punkt")
duplicates = []
for pair in itertools.combinations(feed.articles, 2):
- try:
- result = cosine_sim(*pair)
- if abs(result.item() - 1.0) < 1e-10:
- duplicates.append(pair)
- #print pair[0].id, pair[0].title, pair[0].link
- #print pair[1].id, pair[1].title, pair[1].link
- #print
- except ValueError:
- continue
+ if pair[0].content != "" and pair[0].content == pair[1].content:
+ duplicates.append(pair)
return duplicates \ No newline at end of file
bgstack15