#4 pororo 라벨링 적용

jayleenym · Jan 6, 2022 · fb2f097 · fb2f097
1 parent 6954bf4
commit fb2f097
Showing 1 changed file with 20 additions and 14 deletions.
diff --git a/Kafka/1228/Naver_Producer1228.ipynb b/Kafka/1228/Naver_Producer1228.ipynb
@@ -21,7 +21,21 @@
     "import datetime\n",
     "import pandas as pd\n",
     "import datetime\n",
-    "from tqdm import tqdm"
+    "from tqdm import tqdm\n",
+    "\n",
+    "# 감성사전 처리해서 쌓기\n",
+    "from pororo import Pororo"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# 여기서 오래 걸릴 수 있음\n",
+    "s_nm = Pororo(task = 'sentiment',model = 'brainbert.base.ko.nsmc', lang = 'ko')\n",
+    "s_sh = Pororo(task = 'sentiment', model = 'brainbert.base.ko.shopping', lang = 'ko')"
    ]
   },
   {
@@ -128,26 +142,18 @@
     "                                      'title':title,\n",
     "                                      'views':views,\n",
     "                                      'pos':pos,\n",
-    "                                       'neg':neg\n",
-    "                                                 }\n",
+    "                                      'neg':neg,\n",
+    "                                      'nsmc': s_nm(title, show_probs = True)['positive'],\n",
+    "                                      'shop': s_sh(title, show_prob = True)['positivie']\n",
+    "                            }\n",
     "                # print(put_data)\n",
     "                title_result.append(put_data)\n",
     "                \n",
     "        #if TF:\n",
     "            #break\n",
     "        return title_result\n",
     "        #else:\n",
-    "        print(title_result)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 34,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "for key,value in ACODE.items():\n",
-    "    result = naver_crawl(value)"
+    "        # print(title_result)"
    ]
   },
   {