{"id":204610,"date":"2025-05-29T10:17:53","date_gmt":"2025-05-29T02:17:53","guid":{"rendered":"https:\/\/server.hk\/cnblog\/204610\/"},"modified":"2025-05-29T10:17:53","modified_gmt":"2025-05-29T02:17:53","slug":"%e5%a6%82%e4%bd%95%e9%81%bf%e5%85%8d%e8%af%8d%e7%bb%84%e6%8b%86%e5%88%86%e5%bd%b1%e5%93%8d-tf-idf-%e8%ae%a1%e7%ae%97%ef%bc%9f","status":"publish","type":"post","link":"https:\/\/server.hk\/cnblog\/204610\/","title":{"rendered":"\u5982\u4f55\u907f\u514d\u8bcd\u7ec4\u62c6\u5206\u5f71\u54cd TF-IDF \u8ba1\u7b97\uff1f"},"content":{"rendered":"

<\/b> <\/p>\n

\u5982\u4f55\u907f\u514d\u8bcd\u7ec4\u62c6\u5206\u5f71\u54cd TF-IDF \u8ba1\u7b97\uff1f <\/h1>\n

\u4e00\u5206\u8015\u8018\uff0c\u4e00\u5206\u6536\u83b7\uff01\u65e2\u7136\u90fd\u6253\u5f00\u8fd9\u7bc7\uff0c\u5c31\u575a\u6301\u770b\u4e0b\u53bb\uff0c\u5b66\u4e0b\u53bb\u5427\uff01\u672c\u6587\u4e3b\u8981\u4f1a\u7ed9\u5927\u5bb6\u8bb2\u5230<\/span>\u7b49\u7b49\u77e5\u8bc6\u70b9\uff0c\u5982\u679c\u5927\u5bb6\u5bf9\u672c\u6587\u6709\u597d\u7684\u5efa\u8bae\u6216\u8005\u770b\u5230\u6709\u4e0d\u8db3\u4e4b\u5904\uff0c\u975e\u5e38\u6b22\u8fce\u5927\u5bb6\u79ef\u6781\u63d0\u51fa\uff01\u5728\u540e\u7eed\u6587\u7ae0\u6211\u4f1a\u7ee7\u7eed\u66f4\u65b0\u6587\u7ae0<\/span>\u76f8\u5173\u7684\u5185\u5bb9\uff0c\u5e0c\u671b\u5bf9\u5927\u5bb6\u90fd\u6709\u6240\u5e2e\u52a9\uff01<\/p>\n

<\/p>\n

\u81ea\u5b9a\u4e49 tf-idf \u8ba1\u7b97\uff0c\u907f\u514d\u8bcd\u7ec4\u62c6\u5206<\/strong><\/p>\n

\u5728\u4f7f\u7528 tfidfvectorizer \u8ba1\u7b97 tf-idf \u503c\u65f6\uff0c\u5f53\u6587\u672c\u6570\u636e\u5305\u542b\u8bcd\u7ec4\u65f6\uff0c\u53ef\u80fd\u4f1a\u9047\u5230\u81ea\u52a8\u5206\u8bcd\u7684\u95ee\u9898\uff0c\u5bfc\u81f4\u8f93\u51fa\u7279\u5f81\u5305\u542b\u5206\u62c6\u540e\u7684\u5355\u8bcd\u3002\u4e3a\u4e86\u89e3\u51b3\u8fd9\u4e00\u95ee\u9898\uff0c\u4ee5\u4e0b\u63d0\u4f9b\u4e24\u79cd\u65b9\u6cd5\uff1a<\/p>\n

1. \u8c03\u6574 tfidfvectorizer \u53c2\u6570<\/strong><\/p>\n

\u5982\u679c\u6587\u672c\u6570\u636e\u4e2d\u7684\u8bcd\u7ec4\u7531\u4e0b\u5212\u7ebf\u6216\u5176\u4ed6\u5b57\u7b26\u8fde\u63a5\uff0c\u53ef\u4ee5\u8bbe\u7f6e tfidfvectorizer \u7684 analyzer \u53c2\u6570\u4e3a “word”\uff0c\u4ee5\u7981\u7528\u5206\u8bcd\u529f\u80fd\u3002<\/p>\n

from sklearn.feature_extraction.text import tfidfvectorizer\n\ndocs = [\"this_is_book\", \"this_is_apple\"]\nvectorizer = tfidfvectorizer(analyzer=\"word\", stop_words=\"english\")\ntfidf = vectorizer.fit_transform(docs)\nprint(vectorizer.get_feature_names_out())<\/pre>\n
\u8f93\u51fa\uff1a<\/strong><\/p>\n
['this_is_apple', 'this_is_book']<\/pre>\n
2. \u81ea\u5b9a\u4e49 tf-idf \u8ba1\u7b97<\/strong><\/p>\n
\u5982\u679c\u4f60\u4e0d\u60f3\u4f7f\u7528 tfidfvectorizer\uff0c\u4e5f\u53ef\u4ee5\u81ea\u884c\u7f16\u5199 tf-idf \u8ba1\u7b97\u7a0b\u5e8f\u3002\u4ee5\u4e0b\u662f\u4e00\u4e2a\u793a\u4f8b\u5b9e\u73b0\uff1a<\/p>\n
import math\n\ndef tfidf_custom(docs, vocab):\n  \"\"\"\n  \u81ea\u5b9a\u4e49 TF-IDF \u8ba1\u7b97\n\n  \u53c2\u6570\uff1a\n    docs: \u6587\u6863\u96c6\u5408\n    vocab: \u8bcd\u6c47\u8868\n  \"\"\"\n\n  # \u8ba1\u7b97\u8bcd\u9891\n  tf_dict = {}\n  for doc in docs:\n    for word in doc:\n      if word in tf_dict.keys():\n        tf_dict[word] += 1\n      else:\n        tf_dict[word] = 1\n\n  # \u8ba1\u7b97\u6587\u6863\u9891\u7387\n  df_dict = {}\n  for word in vocab:\n    for doc in docs:\n      if word in doc:\n        if word in df_dict.keys():\n          df_dict[word] += 1\n        else:\n          df_dict[word] = 1\n\n  # \u8ba1\u7b97 TF-IDF \u503c\n  tfidf_dict = {}\n  for word in vocab:\n    tfidf_dict[word] = (tf_dict[word] \/ sum(tf_dict.values())) * math.log(len(docs) \/ df_dict[word])\n\n  return tfidf_dict\n\n# \u6587\u6863\u96c6\u5408\u548c\u8bcd\u6c47\u8868\ndocs = [\"This_is_book\", \"This_is_apple\"]\nvocab = [\"This_is_apple\", \"This_is_book\"]\n\ntfidf_custom(docs, vocab)<\/pre>\n
\u4eca\u5929\u5173\u4e8e\u300a\u5982\u4f55\u907f\u514d\u8bcd\u7ec4\u62c6\u5206\u5f71\u54cd TF-IDF \u8ba1\u7b97\uff1f \u300b\u7684\u5185\u5bb9\u5c31\u4ecb\u7ecd\u5230\u8fd9\u91cc\u4e86\uff0c\u662f\u4e0d\u662f\u5b66\u8d77\u6765\u4e00\u76ee\u4e86\u7136\uff01\u60f3\u8981\u4e86\u89e3\u66f4\u591a\u5173\u4e8e\u7684\u5185\u5bb9\u8bf7\u5173\u6ce8\u516c\u4f17\u53f7\uff01<\/p>\n","protected":false},"excerpt":{"rendered":"
\u5982\u4f55\u907f\u514d\u8bcd\u7ec4\u62c6\u5206\u5f71\u54cd TF-ID...<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[4925],"tags":[],"class_list":["post-204610","post","type-post","status-publish","format-standard","hentry","category-4925"],"_links":{"self":[{"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/posts\/204610","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/comments?post=204610"}],"version-history":[{"count":0,"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/posts\/204610\/revisions"}],"wp:attachment":[{"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/media?parent=204610"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/categories?post=204610"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/tags?post=204610"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}