def __init__(self, lang="spanish"):
"""
Initializes the parameters for specific language
"""
self.languages = ["spanish", "english", "italian", "german"]
self.lang = lang
if self.lang not in self.languages:
raise LangDependencyError("Language not supported: " + lang)
self.stopwords = LangDependency.STOPWORDS_CACHE.get(lang, None)
if self.stopwords is None:
self.stopwords = self.load_stopwords(os.path.join(PATH, "{0}.stopwords".format(lang)))
LangDependency.STOPWORDS_CACHE[lang] = self.stopwords
self.neg_stopwords = LangDependency.NEG_STOPWORDS_CACHE.get(lang, None)
if self.neg_stopwords is None:
self.neg_stopwords = self.load_stopwords(os.path.join(PATH, "{0}.neg.stopwords".format(lang)))
LangDependency.NEG_STOPWORDS_CACHE[lang] = self.neg_stopwords
if self.lang not in SnowballStemmer.languages:
raise LangDependencyError("Language not supported for stemming: " + lang)
if self.lang == "english":
self.stemmer = PorterStemmer()
else:
self.stemmer = SnowballStemmer(self.lang)
评论列表
文章目录