Format: application/x-python-code
⬇ Download File (1.4 KB)
   1  U
   2  
   3  ���`.�@s�ddlZddlZddlmZmZmZddlm	Z	ddl
   4  Z
   5  e
   6  �d�Ze	�
   7  d�Z	e	�dd�e	D��e	�dd	d
   8  ddd
   9  ddddddddddddg�Gdd�d�ZdS)�N)�CountVectorizer�TfidfTransformer�TfidfVectorizer)�	stopwordsZen_core_web_sm�englishcCsg|]}|���qS�)�
  10  capitalize)�.0�srr�)C:\Users\PC\Pytho\Projects\NLP\Process.py�
  11  <listcomp>sr�,�`�?z'sz''z``zn't�(�)�;�'�re�!z--�-�*z're�:c@s0eZdZd
  12  dd�Zdddddd	gefd
  13  d�ZdS)�Process�cCs
  14  ||_dS)N)�m)�selfrrrr�__init__szProcess.__init__ZNNSZNNZADJZJJZJJRZVBcs*t|�}�fdd�tjD�}d�|�}|S)Ncsg|]}|j�kr|j�qSr)Ztag_Zlemma_)r	�token��allowed_posrrrs
  15  z'Process.pre_process.<locals>.<listcomp>� )�nlpr�words�join)�textr rr#�new_textrrr�pre_processs
  16  zProcess.pre_processN)r)�__name__�
  17  __module__�__qualname__rrr'rrrrrs
  18  r)�pandas�pd�numpy�npZsklearn.feature_extraction.textrrrZnltk.corpusrZspacy�loadr"r#�extendrrrrr�<module>s
  19  
  20  .