Assistant.py 文件源码

def fill_tf_idf_shelve(self):
        tf_idf_shelve = shelve.open(self.tf_idf_shelve_file_name, writeback=True)
        if TF not in tf_idf_shelve:
            tf_idf_shelve[TF] = {}
        if DF not in tf_idf_shelve:
            tf_idf_shelve[DF] = {}
        if D not in tf_idf_shelve:
            tf_idf_shelve[D] = 0
        if TF_IDF not in tf_idf_shelve:
            tf_idf_shelve[TF_IDF] = {}
        if CENTROID not in tf_idf_shelve:
            tf_idf_shelve[CENTROID] = {}

        for action,trigger_txt in self.trigger_dict.iteritems():
            if action not in tf_idf_shelve[TF].keys():
                trigger = self.tokenize_text(trigger_txt)
                tf_idf_shelve[TF][action] = Counter(trigger)
                for word in unique(trigger):
                    if word not in tf_idf_shelve[DF].keys():
                        tf_idf_shelve[DF][word] = 0
                    tf_idf_shelve[DF][word] += 1
        tf_idf_shelve[D] = len(tf_idf_shelve[TF])
        tf_idf_shelve.close()
        self.compute_tf_idf()
        self.compute_centroids()