{"id":112444,"date":"2026-10-03T17:16:52","date_gmt":"2026-10-03T09:16:52","guid":{"rendered":"https:\/\/www.wsisp.com\/helps\/112444.html"},"modified":"2026-10-03T17:16:52","modified_gmt":"2026-10-03T09:16:52","slug":"nlp%e5%9f%ba%e7%a1%80%e5%88%b0%e9%ab%98%e7%ba%a702%ef%bc%9a%e8%af%8d%e8%a2%8b%e6%a8%a1%e5%9e%8b%e3%80%81tf-idf-%e4%b8%8e%e6%96%87%e6%9c%ac%e8%a1%a8%e7%a4%ba","status":"publish","type":"post","link":"https:\/\/www.wsisp.com\/helps\/112444.html","title":{"rendered":"NLP\u57fa\u7840\u5230\u9ad8\u7ea702\uff1a\u8bcd\u888b\u6a21\u578b\u3001TF-IDF \u4e0e\u6587\u672c\u8868\u793a"},"content":{"rendered":"

\u8bcd\u888b\u6a21\u578b\u3001TF-IDF \u4e0e\u6587\u672c\u8868\u793a<\/h2>\n

\u5148\u8ba1\u6570，\u540e\u601d\u8003\u3002TF-IDF \u5728\u5b9a\u4e49\u660e\u786e\u7684\u4efb\u52a1\u4e0a\u4f9d\u7136\u80dc\u8fc7\u5d4c\u5165\u6a21\u578b，\u5373\u4fbf\u5728 2026 \u5e74\u3002<\/p>\n

\u7c7b\u578b： \u6784\u5efa
\n\u8bed\u8a00： Python
\n\u524d\u7f6e\u6761\u4ef6： Phase 5 \u00b7 01 (\u6587\u672c\u5904\u7406)，Phase 2 \u00b7 02 (\u4ece\u96f6\u5b9e\u73b0\u7ebf\u6027\u56de\u5f52)
\n\u7528\u65f6： ~75 \u5206\u949f<\/p>\n

\u95ee\u9898\u6240\u5728<\/h3>\n

\u6a21\u578b\u9700\u8981\u6570\u5b57，\u800c\u4f60\u53ea\u6709\u5b57\u7b26\u4e32\u3002<\/p>\n

\u6bcf\u4e2a NLP \u7ba1\u9053\u90fd\u5fc5\u987b\u56de\u7b54\u540c\u4e00\u4e2a\u95ee\u9898：\u5982\u4f55\u5c06\u53d8\u957f\u7684 token \u6d41\u8f6c\u5316\u4e3a\u5206\u7c7b\u5668\u53ef\u4ee5\u6d88\u8d39\u7684\u56fa\u5b9a\u5927\u5c0f\u5411\u91cf\u3002\u8fd9\u4e2a\u9886\u57df\u7ed9\u51fa\u7684\u7b2c\u4e00\u4e2a\u7b54\u6848\u662f\u80fd\u7528\u7684\u6700\u7b28\u65b9\u6848\u2014\u2014\u6570\u8bcd\u9891，\u6784\u9020\u6210\u5411\u91cf\u3002<\/p>\n

\u8fd9\u4e2a\u5411\u91cf\u627f\u8f7d\u4e86\u6bd4\u4efb\u4f55\u5d4c\u5165\u6a21\u578b\u90fd\u591a\u7684\u751f\u4ea7\u7ea7 NLP \u5e94\u7528\u3002\u5783\u573e\u90ae\u4ef6\u8fc7\u6ee4\u5668\u3001\u4e3b\u9898\u5206\u7c7b\u5668\u3001\u65e5\u5fd7\u5f02\u5e38\u68c0\u6d4b\u3001\u641c\u7d22\u6392\u5e8f（\u5728 BM25 \u4e4b\u524d）\u3001\u7b2c\u4e00\u6ce2\u60c5\u611f\u5206\u6790\u3001\u5b66\u672f\u754c NLP \u57fa\u51c6\u6d4b\u8bd5\u7684\u5934\u5341\u5e74\u30022026 \u5e74\u7684\u4ece\u4e1a\u8005\u5728\u7a84\u5206\u7c7b\u4efb\u52a1\u4e0a\u4ecd\u7136\u4f1a\u9996\u5148\u9009\u62e9\u5b83\u3002\u5b83\u5feb\u901f\u3001\u53ef\u89e3\u91ca，\u5728\u8bcd\u7684\u51fa\u73b0\u4e0e\u5426\u662f\u5173\u952e\u56e0\u7d20\u7684\u4efb\u52a1\u4e0a，\u5f80\u5f80\u4e0e 4 \u4ebf\u53c2\u6570\u7684\u5d4c\u5165\u6a21\u578b\u96be\u4ee5\u533a\u5206\u3002<\/p>\n

\u672c\u8bfe\u4ece\u96f6\u6784\u5efa\u8bcd\u888b\u6a21\u578b\u548c TF-IDF，\u7136\u540e\u5c55\u793a scikit-learn \u5982\u4f55\u7528\u4e09\u884c\u4ee3\u7801\u5b8c\u6210\u540c\u6837\u7684\u5de5\u4f5c，\u6700\u540e\u6307\u51fa\u8ba9\u4f60\u8f6c\u5411\u5d4c\u5165\u6a21\u578b\u7684\u5931\u8d25\u6a21\u5f0f\u3002<\/p>\n

\u6838\u5fc3\u6982\u5ff5<\/h3>\n

\u8bcd\u888b\u6a21\u578b (Bag of Words, BoW) \u4e22\u5f03\u4e86\u8bcd\u5e8f\u3002\u5bf9\u6bcf\u4e2a\u6587\u6863，\u7edf\u8ba1\u6bcf\u4e2a\u8bcd\u6c47\u8868\u4e2d\u8bcd\u51fa\u73b0\u7684\u6b21\u6570\u3002\u5411\u91cf\u957f\u5ea6\u7b49\u4e8e\u8bcd\u6c47\u8868\u5927\u5c0f，\u4f4d\u7f6e i \u5b58\u653e\u8bcd i \u7684\u8ba1\u6570\u3002<\/p>\n

TF-IDF \u5bf9 BoW \u8fdb\u884c\u91cd\u65b0\u52a0\u6743\u3002\u51fa\u73b0\u5728\u6bcf\u4e2a\u6587\u6863\u4e2d\u7684\u8bcd\u4fe1\u606f\u91cf\u4f4e，\u6240\u4ee5\u7f29\u5c0f\u5176\u6743\u91cd；\u5728\u6574\u4e2a\u8bed\u6599\u5e93\u4e2d\u7a00\u5c11\u4f46\u5728\u5355\u4e2a\u6587\u6863\u4e2d\u9891\u7e41\u51fa\u73b0\u7684\u8bcd\u662f\u4fe1\u53f7，\u6240\u4ee5\u653e\u5927\u5176\u6743\u91cd\u3002<\/p>\n

TF-IDF(w, d) = TF(w, d) * IDF(w)
\n = count(w in d) \/ |d| * log(N \/ df(w))<\/p>\n

\u5176\u4e2d TF \u662f\u6587\u6863\u4e2d\u7684\u8bcd\u9891，df \u662f\u6587\u6863\u9891\u7387（\u5305\u542b\u8be5\u8bcd\u7684\u6587\u6863\u6570\u91cf），N \u662f\u6587\u6863\u603b\u6570\u3002log \u4f7f\u65e0\u5904\u4e0d\u5728\u7684\u8bcd\u7684\u6743\u91cd\u4fdd\u6301\u6709\u754c\u3002<\/p>\n

\u5173\u952e\u7279\u6027：\u4e24\u8005\u90fd\u4ea7\u751f\u5177\u6709\u53ef\u89e3\u91ca\u8f74\u7684\u7a00\u758f\u5411\u91cf\u3002\u4f60\u53ef\u4ee5\u67e5\u770b\u8bad\u7ec3\u597d\u7684\u5206\u7c7b\u5668\u7684\u6743\u91cd，\u8bfb\u51fa\u54ea\u4e9b\u8bcd\u5c06\u6587\u6863\u63a8\u5411\u5404\u4e2a\u7c7b\u522b\u3002\u4f60\u65e0\u6cd5\u7528 768 \u7ef4\u7684 BERT \u5d4c\u5165\u505a\u5230\u8fd9\u4e00\u70b9\u3002<\/p>\n

\u52a8\u624b\u6784\u5efa<\/h3>\n

\u6b65\u9aa4 1：\u6784\u5efa\u8bcd\u6c47\u8868<\/h4>\n

def<\/span> build_vocab<\/span>(<\/span>docs)<\/span>:<\/span>
\n vocab =<\/span> {<\/span>}<\/span>
\n for<\/span> doc in<\/span> docs:<\/span>
\n for<\/span> token in<\/span> doc:<\/span>
\n if<\/span> token not<\/span> in<\/span> vocab:<\/span>
\n vocab[<\/span>token]<\/span> =<\/span> len<\/span>(<\/span>vocab)<\/span>
\n return<\/span> vocab<\/p>\n

\u8f93\u5165：\u5df2\u5206\u8bcd\u7684\u6587\u6863\u5217\u8868（\u4efb\u4f55\u8bcd\u7ea7\u522b\u5206\u8bcd\u5668\u90fd\u53ef\u4ee5；\u672c\u8bfe\u7684 code\/main.py \u4f7f\u7528\u7b80\u5316\u7684\u7eaf\u5c0f\u5199\u7248\u672c）\u3002\u8f93\u51fa：{word: index} \u5b57\u5178\u3002\u7a33\u5b9a\u7684\u63d2\u5165\u987a\u5e8f\u610f\u5473\u7740\u7d22\u5f15 0 \u5bf9\u5e94\u7b2c\u4e00\u4e2a\u6587\u6863\u4e2d\u7b2c\u4e00\u4e2a\u51fa\u73b0\u7684\u8bcd\u3002\u60ef\u4f8b\u5404\u4e0d\u76f8\u540c；scikit-learn \u6309\u5b57\u6bcd\u6392\u5e8f\u3002<\/p>\n

\u6b65\u9aa4 2：\u8bcd\u888b\u6a21\u578b<\/h4>\n

def<\/span> bag_of_words<\/span>(<\/span>docs,<\/span> vocab)<\/span>:<\/span>
\n matrix =<\/span> [<\/span>[<\/span>0<\/span>]<\/span> *<\/span> len<\/span>(<\/span>vocab)<\/span> for<\/span> _ in<\/span> docs]<\/span>
\n for<\/span> i,<\/span> doc in<\/span> enumerate<\/span>(<\/span>docs)<\/span>:<\/span>
\n for<\/span> token in<\/span> doc:<\/span>
\n if<\/span> token in<\/span> vocab:<\/span>
\n matrix[<\/span>i]<\/span>[<\/span>vocab[<\/span>token]<\/span>]<\/span> +=<\/span> 1<\/span>
\n return<\/span> matrix<\/p>\n

>><\/span>><\/span> docs =<\/span> [<\/span>[<\/span>"cat"<\/span>,<\/span> "sat"<\/span>,<\/span> "on"<\/span>,<\/span> "mat"<\/span>]<\/span>,<\/span> [<\/span>"cat"<\/span>,<\/span> "cat"<\/span>,<\/span> "ran"<\/span>]<\/span>]<\/span>
\n>><\/span>><\/span> vocab =<\/span> build_vocab(<\/span>docs)<\/span>
\n>><\/span>><\/span> bag_of_words(<\/span>docs,<\/span> vocab)<\/span>
\n[<\/span>[<\/span>1<\/span>,<\/span> 1<\/span>,<\/span> 1<\/span>,<\/span> 1<\/span>,<\/span> 0<\/span>]<\/span>,<\/span> [<\/span>2<\/span>,<\/span> 0<\/span>,<\/span> 0<\/span>,<\/span> 0<\/span>,<\/span> 1<\/span>]<\/span>]<\/span><\/p>\n

\u884c\u662f\u6587\u6863，\u5217\u662f\u8bcd\u6c47\u7d22\u5f15\u3002\u6761\u76ee [i][j] \u8868\u793a"\u8bcd j \u5728\u6587\u6863 i \u4e2d\u51fa\u73b0\u4e86\u591a\u5c11\u6b21"\u3002\u6587\u6863 1 \u4e2d cat \u51fa\u73b0\u4e86\u4e24\u6b21，\u6240\u4ee5\u662f 2\u3002\u6587\u6863 0 \u4e2d ran \u6ca1\u51fa\u73b0，\u6240\u4ee5\u662f 0\u3002<\/p>\n

\u6b65\u9aa4 3：\u8bcd\u9891\u4e0e\u6587\u6863\u9891\u7387<\/h4>\n

import<\/span> math<\/p>\n

def<\/span> term_frequency<\/span>(<\/span>doc_bow,<\/span> doc_length)<\/span>:<\/span>
\n return<\/span> [<\/span>c \/<\/span> doc_length if<\/span> doc_length else<\/span> 0<\/span> for<\/span> c in<\/span> doc_bow]<\/span><\/p>\n

def<\/span> document_frequency<\/span>(<\/span>bow_matrix)<\/span>:<\/span>
\n df =<\/span> [<\/span>0<\/span>]<\/span> *<\/span> len<\/span>(<\/span>bow_matrix[<\/span>0<\/span>]<\/span>)<\/span>
\n for<\/span> row in<\/span> bow_matrix:<\/span>
\n for<\/span> j,<\/span> count in<\/span> enumerate<\/span>(<\/span>row)<\/span>:<\/span>
\n if<\/span> count ><\/span> 0<\/span>:<\/span>
\n df[<\/span>j]<\/span> +=<\/span> 1<\/span>
\n return<\/span> df<\/p>\n

def<\/span> inverse_document_frequency<\/span>(<\/span>df,<\/span> n_docs)<\/span>:<\/span>
\n return<\/span> [<\/span>math.<\/span>log(<\/span>(<\/span>n_docs +<\/span> 1<\/span>)<\/span> \/<\/span> (<\/span>d +<\/span> 1<\/span>)<\/span>)<\/span> +<\/span> 1<\/span> for<\/span> d in<\/span> df]<\/span><\/p>\n

\u4e24\u4e2a\u503c\u5f97\u6307\u51fa\u7684\u5e73\u6ed1\u6280\u5de7\u3002(n+1)\/(d+1) \u907f\u514d log(x\/0)\u3002\u5c3e\u90e8\u7684 +1 \u786e\u4fdd\u51fa\u73b0\u5728\u6bcf\u4e2a\u6587\u6863\u4e2d\u7684\u8bcd IDF \u4ecd\u4e3a 1（\u800c\u975e 0），\u4e0e scikit-learn \u7684\u9ed8\u8ba4\u884c\u4e3a\u4e00\u81f4\u3002\u5176\u4ed6\u5b9e\u73b0\u4f7f\u7528\u539f\u59cb\u7684 log(N\/df)\u3002\u4e24\u79cd\u65b9\u5f0f\u90fd\u53ef\u4ee5；\u5e73\u6ed1\u7248\u672c\u66f4\u53cb\u597d\u3002<\/p>\n

\u6b65\u9aa4 4：TF-IDF<\/h4>\n

def<\/span> tfidf<\/span>(<\/span>bow_matrix)<\/span>:<\/span>
\n n_docs =<\/span> len<\/span>(<\/span>bow_matrix)<\/span>
\n df =<\/span> document_frequency(<\/span>bow_matrix)<\/span>
\n idf =<\/span> inverse_document_frequency(<\/span>df,<\/span> n_docs)<\/span>
\n out =<\/span> [<\/span>]<\/span>
\n for<\/span> row in<\/span> bow_matrix:<\/span>
\n length =<\/span> sum<\/span>(<\/span>row)<\/span>
\n tf =<\/span> term_frequency(<\/span>row,<\/span> length)<\/span>
\n out.<\/span>append(<\/span>[<\/span>tf_j *<\/span> idf_j for<\/span> tf_j,<\/span> idf_j in<\/span> zip<\/span>(<\/span>tf,<\/span> idf)<\/span>]<\/span>)<\/span>
\n return<\/span> out<\/p>\n

>><\/span>><\/span> docs =<\/span> [<\/span>
\n.<\/span>.<\/span>.<\/span> [<\/span>"the"<\/span>,<\/span> "cat"<\/span>,<\/span> "sat"<\/span>]<\/span>,<\/span>
\n.<\/span>.<\/span>.<\/span> [<\/span>"the"<\/span>,<\/span> "dog"<\/span>,<\/span> "sat"<\/span>]<\/span>,<\/span>
\n.<\/span>.<\/span>.<\/span> [<\/span>"the"<\/span>,<\/span> "cat"<\/span>,<\/span> "ran"<\/span>]<\/span>,<\/span>
\n.<\/span>.<\/span>.<\/span> ]<\/span>
\n>><\/span>><\/span> vocab =<\/span> build_vocab(<\/span>docs)<\/span>
\n>><\/span>><\/span> bow =<\/span> bag_of_words(<\/span>docs,<\/span> vocab)<\/span>
\n>><\/span>><\/span> tfidf(<\/span>bow)<\/span><\/p>\n

\u4e09\u4e2a\u6587\u6863，\u4e94\u4e2a\u8bcd\u6c47\u8bcd（the\u3001cat\u3001sat\u3001dog\u3001ran）\u3002the \u51fa\u73b0\u5728\u5168\u90e8\u4e09\u4e2a\u6587\u6863\u4e2d，\u6240\u4ee5 IDF \u4f4e\u3002dog \u53ea\u51fa\u73b0\u5728\u4e00\u4e2a\u6587\u6863\u4e2d，\u6240\u4ee5 IDF \u9ad8\u3002\u5411\u91cf\u662f\u7a00\u758f\u7684（\u5927\u591a\u6570\u6761\u76ee\u5f88\u5c0f），\u533a\u5206\u6027\u5f3a\u7684\u8bcd\u51f8\u663e\u51fa\u6765\u3002<\/p>\n

\u6b65\u9aa4 5：L2 \u5f52\u4e00\u5316<\/h4>\n

def<\/span> l2_normalize<\/span>(<\/span>matrix)<\/span>:<\/span>
\n out =<\/span> [<\/span>]<\/span>
\n for<\/span> row in<\/span> matrix:<\/span>
\n norm =<\/span> math.<\/span>sqrt(<\/span>sum<\/span>(<\/span>x *<\/span> x for<\/span> x in<\/span> row)<\/span>)<\/span>
\n out.<\/span>append(<\/span>[<\/span>x \/<\/span> norm if<\/span> norm else<\/span> 0<\/span> for<\/span> x in<\/span> row]<\/span>)<\/span>
\n return<\/span> out<\/p>\n

\u6ca1\u6709\u5f52\u4e00\u5316，\u8f83\u957f\u7684\u6587\u6863\u4f1a\u5f97\u5230\u66f4\u5927\u7684\u5411\u91cf，\u5728\u76f8\u4f3c\u5ea6\u8ba1\u7b97\u4e2d\u5360\u4e3b\u5bfc\u3002L2 \u5f52\u4e00\u5316\u5c06\u6bcf\u4e2a\u6587\u6863\u6620\u5c04\u5230\u5355\u4f4d\u8d85\u7403\u9762\u4e0a\u3002\u884c\u4e4b\u95f4\u7684\u4f59\u5f26\u76f8\u4f3c\u5ea6\u73b0\u5728\u5c31\u662f\u70b9\u79ef\u3002<\/p>\n

\u5b9e\u9645\u5e94\u7528<\/h3>\n

scikit-learn \u63d0\u4f9b\u4e86\u751f\u4ea7\u7ea7\u7248\u672c\u3002<\/p>\n

from<\/span> sklearn.<\/span>feature_extraction.<\/span>text import<\/span> CountVectorizer,<\/span> TfidfVectorizer<\/p>\n

docs =<\/span> [<\/span>"the cat sat on the mat"<\/span>,<\/span> "the dog sat on the mat"<\/span>,<\/span> "the cat ran"<\/span>]<\/span><\/p>\n

bow_vectorizer =<\/span> CountVectorizer(<\/span>)<\/span>
\nbow =<\/span> bow_vectorizer.<\/span>fit_transform(<\/span>docs)<\/span>
\nprint<\/span>(<\/span>bow_vectorizer.<\/span>get_feature_names_out(<\/span>)<\/span>)<\/span>
\nprint<\/span>(<\/span>bow.<\/span>toarray(<\/span>)<\/span>)<\/span><\/p>\n

tfidf_vectorizer =<\/span> TfidfVectorizer(<\/span>)<\/span>
\ntfidf =<\/span> tfidf_vectorizer.<\/span>fit_transform(<\/span>docs)<\/span>
\nprint<\/span>(<\/span>tfidf.<\/span>toarray(<\/span>)<\/span>.<\/span>round<\/span>(<\/span>3<\/span>)<\/span>)<\/span><\/p>\n

CountVectorizer \u4e00\u6b21\u8c03\u7528\u5b8c\u6210\u5206\u8bcd\u3001\u8bcd\u6c47\u6784\u5efa\u548c BoW\u3002TfidfVectorizer \u5728\u6b64\u57fa\u7840\u4e0a\u52a0\u5165 IDF \u52a0\u6743\u548c L2 \u5f52\u4e00\u5316\u3002\u4e24\u8005\u90fd\u8fd4\u56de\u7a00\u758f\u77e9\u9635\u3002\u5bf9\u4e8e 10 \u4e07\u6587\u6863，\u5bc6\u96c6\u7248\u672c\u65e0\u6cd5\u653e\u5165\u5185\u5b58；\u4fdd\u6301\u7a00\u758f\u76f4\u5230\u5206\u7c7b\u5668\u8981\u6c42\u5bc6\u96c6\u683c\u5f0f\u3002<\/p>\n

\u6539\u53d8\u4e00\u5207\u7684\u5173\u952e\u53c2\u6570：<\/p>\n\n\u53c2\u6570\u6548\u679c<\/tr>\n\n\n\n\n\n\n
ngram_range=(1, 2)<\/td>\n\u5305\u542b bigram\u3002\u901a\u5e38\u63d0\u5347\u5206\u7c7b\u6548\u679c\u3002<\/td>\n<\/tr>\n
min_df=2<\/td>\n\u4e22\u5f03\u51fa\u73b0\u5728\u5c11\u4e8e 2 \u4e2a\u6587\u6863\u4e2d\u7684\u8bcd\u3002\u5728\u566a\u58f0\u6570\u636e\u4e0a\u7f29\u51cf\u8bcd\u6c47\u8868\u3002<\/td>\n<\/tr>\n
max_df=0.95<\/td>\n\u4e22\u5f03\u51fa\u73b0\u5728\u8d85\u8fc7 95% \u6587\u6863\u4e2d\u7684\u8bcd\u3002\u65e0\u9700\u786c\u7f16\u7801\u5217\u8868\u5373\u53ef\u8fd1\u4f3c\u79fb\u9664\u505c\u7528\u8bcd\u3002<\/td>\n<\/tr>\n
stop_words="english"<\/td>\nscikit-learn \u5185\u7f6e\u505c\u7528\u8bcd\u8868\u3002\u53d6\u51b3\u4e8e\u4efb\u52a1\u2014\u2014\u60c5\u611f\u5206\u6790\u4e0d\u5e94\u4e22\u5f03\u5426\u5b9a\u8bcd\u3002<\/td>\n<\/tr>\n
sublinear_tf=True<\/td>\n\u4f7f\u7528 1 + log(tf) \u4ee3\u66ff\u539f\u59cb tf\u3002\u5f53\u4e00\u4e2a\u8bcd\u5728\u67d0\u4e2a\u6587\u6863\u4e2d\u91cd\u590d\u591a\u6b21\u65f6\u6709\u5e2e\u52a9\u3002<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

TF-IDF \u4ecd\u7136\u80dc\u51fa\u7684\u573a\u666f（\u622a\u81f3 2026 \u5e74）<\/h4>\n
    \n
  • \u5783\u573e\u90ae\u4ef6\u68c0\u6d4b\u3001\u4e3b\u9898\u6807\u6ce8\u3001\u65e5\u5fd7\u5f02\u5e38\u6807\u8bb0\u3002\u8bcd\u7684\u51fa\u73b0\u662f\u5173\u952e；\u8bed\u4e49\u7ec6\u5fae\u5dee\u522b\u4e0d\u91cd\u8981\u3002<\/li>\n
  • \u4f4e\u6570\u636e\u573a\u666f（\u6570\u767e\u4e2a\u6807\u6ce8\u6837\u672c）\u3002TF-IDF \u52a0\u903b\u8f91\u56de\u5f52\u6ca1\u6709\u9884\u8bad\u7ec3\u6210\u672c\u3002<\/li>\n
  • \u4efb\u4f55\u5ef6\u8fdf\u654f\u611f\u7684\u573a\u666f\u3002TF-IDF \u52a0\u7ebf\u6027\u6a21\u578b\u5728\u5fae\u79d2\u7ea7\u54cd\u5e94\u3002\u901a\u8fc7 Transformer \u5d4c\u5165\u6587\u6863\u9700\u8981 10-100ms\u3002<\/li>\n
  • \u9700\u8981\u89e3\u91ca\u9884\u6d4b\u7ed3\u679c\u7684\u7cfb\u7edf\u3002\u68c0\u67e5\u5206\u7c7b\u5668\u7cfb\u6570，\u6392\u540d\u9760\u524d\u7684\u6b63\u5411\u8bcd\u5c31\u662f\u539f\u56e0\u3002<\/li>\n<\/ul>\n

    TF-IDF \u5931\u8d25\u7684\u573a\u666f<\/h4>\n

    \u8bed\u4e49\u76f2\u533a\u5931\u8d25\u3002\u770b\u8fd9\u4e24\u4e2a\u6587\u6863：<\/p>\n

      \n
    • \u201cThe movie was not good at all.\u201d<\/li>\n
    • \u201cThe movie was excellent.\u201d<\/li>\n<\/ul>\n

      \u4e00\u6761\u662f\u8d1f\u9762\u8bc4\u4ef7，\u4e00\u6761\u662f\u6b63\u9762\u8bc4\u4ef7\u3002\u5b83\u4eec\u7684 TF-IDF \u91cd\u53e0\u90e8\u5206\u53ea\u6709 {the, movie, was}\u3002\u8bcd\u888b\u5206\u7c7b\u5668\u5fc5\u987b\u8bb0\u4f4f not \u51fa\u73b0\u5728 good \u9644\u8fd1\u4f1a\u7ffb\u8f6c\u6807\u7b7e\u3002\u5728\u8db3\u591f\u591a\u7684\u6570\u636e\u4e0a\u5b83\u53ef\u4ee5\u5b66\u5230，\u4f46\u6c38\u8fdc\u4e0d\u5982\u7406\u89e3\u53e5\u6cd5\u7684\u6a21\u578b\u6765\u5f97\u4f18\u96c5\u3002<\/p>\n

      \u53e6\u4e00\u4e2a\u5931\u8d25：\u63a8\u7406\u65f6\u7684\u8bcd\u6c47\u5916 (out-of-vocabulary) \u8bcd\u3002\u5728 IMDb \u8bc4\u8bba\u4e0a\u8bad\u7ec3\u7684 BoW \u6a21\u578b\u5bf9 Zoomer-approved \u8fd9\u79cd\u8bad\u7ec3\u4e2d\u4ece\u672a\u51fa\u73b0\u7684 token \u5b8c\u5168\u65e0\u80fd\u4e3a\u529b\u3002\u5b50\u8bcd\u5d4c\u5165（\u7b2c 04 \u8bfe）\u53ef\u4ee5\u5904\u7406\u8fd9\u4e2a\u95ee\u9898\u3002TF-IDF \u4e0d\u884c\u3002<\/p>\n

      \u6df7\u5408\u65b9\u6848：TF-IDF \u52a0\u6743\u5d4c\u5165<\/h4>\n

      2026 \u5e74\u4e2d\u7b49\u6570\u636e\u91cf\u5206\u7c7b\u7684\u5b9e\u7528\u9ed8\u8ba4\u65b9\u6848：\u5c06 TF-IDF \u6743\u91cd\u4f5c\u4e3a\u8bcd\u5d4c\u5165\u4e0a\u7684\u6ce8\u610f\u529b\u6743\u91cd\u3002<\/p>\n

      def<\/span> tfidf_weighted_embedding<\/span>(<\/span>doc,<\/span> tfidf_scores,<\/span> embedding_table,<\/span> dim)<\/span>:<\/span>
      \n vec =<\/span> [<\/span>0.0<\/span>]<\/span> *<\/span> dim
      \n total_weight =<\/span> 0.0<\/span>
      \n for<\/span> token in<\/span> doc:<\/span>
      \n if<\/span> token not<\/span> in<\/span> embedding_table or<\/span> token not<\/span> in<\/span> tfidf_scores:<\/span>
      \n continue<\/span>
      \n weight =<\/span> tfidf_scores[<\/span>token]<\/span>
      \n emb =<\/span> embedding_table[<\/span>token]<\/span>
      \n for<\/span> i in<\/span> range<\/span>(<\/span>dim)<\/span>:<\/span>
      \n vec[<\/span>i]<\/span> +=<\/span> weight *<\/span> emb[<\/span>i]<\/span>
      \n total_weight +=<\/span> weight
      \n if<\/span> total_weight ==<\/span> 0<\/span>:<\/span>
      \n return<\/span> vec
      \n return<\/span> [<\/span>v \/<\/span> total_weight for<\/span> v in<\/span> vec]<\/span><\/p>\n

      \u4f60\u4ece\u5d4c\u5165\u83b7\u5f97\u8bed\u4e49\u80fd\u529b，\u4ece TF-IDF \u83b7\u5f97\u7a00\u6709\u8bcd\u7684\u5f3a\u8c03\u3002\u5206\u7c7b\u5668\u5728\u6c60\u5316\u540e\u7684\u5411\u91cf\u4e0a\u8bad\u7ec3\u3002\u5728\u5927\u7ea6 5 \u4e07\u4e2a\u6807\u6ce8\u6837\u672c\u4ee5\u4e0b\u7684\u60c5\u611f\u5206\u6790\u3001\u4e3b\u9898\u5206\u7c7b\u548c\u610f\u56fe\u5206\u7c7b\u4efb\u52a1\u4e0a，\u8fd9\u79cd\u65b9\u6848\u4f18\u4e8e\u5355\u72ec\u4f7f\u7528\u4efb\u4e00\u65b9\u6cd5\u3002<\/p>\n

      \u4ea4\u4ed8<\/h3>\n

      \u4fdd\u5b58\u4e3a outputs\/prompt-vectorization-picker.md：<\/p>\n

      —
      \nname: vectorization-picker
      \ndescription: Given a text-classification task, recommend BoW, TF-IDF, embeddings, or a hybrid.
      \nphase: 5
      \nlesson: 02
      \n—<\/p>\n

      You recommend a text-vectorization strategy. Given a task description, output:<\/p>\n

      1. Representation (BoW, TF-IDF, transformer embeddings, or a hybrid). Explain why in one sentence.
      \n2. Specific vectorizer configuration. Name the library. Quote the arguments (`ngram_range`, `min_df`, `max_df`, `sublinear_tf`, `stop_words`).
      \n3. One failure mode to test before shipping.<\/p>\n

      Refuse to recommend embeddings when the user has under 500 labeled examples unless they show evidence of semantic failure in a TF-IDF baseline. Refuse to remove stopwords for sentiment analysis (negations carry signal). Flag class imbalance as needing more than a vectorizer change.<\/p>\n

      Example input: "Classifying 30k customer support tickets into 12 categories. Most tickets are 2-3 sentences. English only. Need explainability for audit logs."<\/p>\n

      Example output:<\/p>\n

      – Representation: TF-IDF. 30k examples is not small; explainability requirement rules out dense embeddings.
      \n– Config: `TfidfVectorizer(ngram_range=(1, 2), min_df=3, max_df=0.95, sublinear_tf=True, stop_words=None)`. Keep stopwords because category keywords sometimes are stopwords ("not working" vs "working").
      \n– Failure to test: verify `min_df=3` does not drop rare category keywords. Run `get_feature_names_out` filtered by class and eyeball.<\/p>\n

      \u7ec3\u4e60<\/h3>\n
    • \u7b80\u5355\u3002 \u5728 L2 \u5f52\u4e00\u5316\u7684 TF-IDF \u8f93\u51fa\u4e0a\u5b9e\u73b0 cosine_similarity(doc_vec_a, doc_vec_b)\u3002\u9a8c\u8bc1\u76f8\u540c\u6587\u6863\u7684\u5f97\u5206\u4e3a 1.0，\u8bcd\u6c47\u5b8c\u5168\u4e0d\u91cd\u53e0\u7684\u6587\u6863\u5f97\u5206\u4e3a 0.0\u3002<\/li>\n
    • \u4e2d\u7b49\u3002 \u4e3a bag_of_words \u6dfb\u52a0 n-gram \u652f\u6301\u3002\u53c2\u6570 n \u751f\u6210 n-gram \u7684\u8ba1\u6570\u3002\u6d4b\u8bd5 n=2 \u5728 ["the", "cat", "sat"] \u4e0a\u4ea7\u751f bigram \u8ba1\u6570 ["the cat", "cat sat"]\u3002<\/li>\n
    • \u56f0\u96be\u3002 \u4f7f\u7528 GloVe 100d \u5411\u91cf\u6784\u5efa\u4e0a\u8ff0 TF-IDF \u52a0\u6743\u5d4c\u5165\u6df7\u5408\u65b9\u6848（\u4e0b\u8f7d\u4e00\u6b21\u5e76\u7f13\u5b58）\u3002\u5728 20 Newsgroups \u6570\u636e\u96c6\u4e0a\u5bf9\u6bd4\u7eaf TF-IDF \u548c\u7eaf\u5e73\u5747\u6c60\u5316\u5d4c\u5165\u7684\u5206\u7c7b\u51c6\u786e\u7387\u3002\u62a5\u544a\u54ea\u79cd\u65b9\u6cd5\u5728\u4f55\u5904\u80dc\u51fa\u3002<\/li>\n

      \u5173\u952e\u672f\u8bed<\/h3>\n\n\u672f\u8bed\u4eba\u4eec\u7684\u8bf4\u6cd5\u5b9e\u9645\u542b\u4e49<\/tr>\n\n\n\n\n\n\n\n
      BoW<\/td>\n\u8bcd\u9891\u5411\u91cf<\/td>\n\u4e00\u4e2a\u6587\u6863\u4e2d\u8bcd\u6c47\u8868\u8bcd\u7684\u8ba1\u6570\u3002\u4e22\u5f03\u4e86\u8bcd\u5e8f\u3002<\/td>\n<\/tr>\n
      TF<\/td>\n\u8bcd\u9891<\/td>\n\u4e00\u4e2a\u8bcd\u5728\u6587\u6863\u4e2d\u7684\u51fa\u73b0\u6b21\u6570，\u53ef\u9009\u62e9\u6309\u6587\u6863\u957f\u5ea6\u5f52\u4e00\u5316\u3002<\/td>\n<\/tr>\n
      DF<\/td>\n\u6587\u6863\u9891\u7387<\/td>\n\u81f3\u5c11\u5305\u542b\u8be5\u8bcd\u4e00\u6b21\u7684\u6587\u6863\u6570\u91cf\u3002<\/td>\n<\/tr>\n
      IDF<\/td>\n\u9006\u6587\u6863\u9891\u7387<\/td>\n\u7ecf\u5e73\u6ed1\u7684 log(N \/ df)\u3002\u964d\u4f4e\u51fa\u73b0\u5728\u6240\u6709\u5730\u65b9\u7684\u8bcd\u7684\u6743\u91cd\u3002<\/td>\n<\/tr>\n
      \u7a00\u758f\u5411\u91cf<\/td>\n\u5927\u90e8\u5206\u4e3a\u96f6<\/td>\n\u8bcd\u6c47\u8868\u901a\u5e38\u6709 1 \u4e07\u5230 10 \u4e07\u4e2a\u8bcd；\u5927\u591a\u6570\u5728\u7ed9\u5b9a\u6587\u6863\u4e2d\u4e0d\u51fa\u73b0\u3002<\/td>\n<\/tr>\n
      \u4f59\u5f26\u76f8\u4f3c\u5ea6<\/td>\n\u5411\u91cf\u5939\u89d2<\/td>\nL2 \u5f52\u4e00\u5316\u5411\u91cf\u7684\u70b9\u79ef\u30021 \u8868\u793a\u76f8\u540c，0 \u8868\u793a\u6b63\u4ea4\u3002<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n","protected":false},"excerpt":{"rendered":"

      \u8bcd\u888b\u6a21\u578b\u3001TF-IDF \u4e0e\u6587\u672c\u8868\u793a\u5148\u8ba1\u6570，\u540e\u601d\u8003\u3002TF-IDF \u5728\u5b9a\u4e49\u660e\u786e\u7684\u4efb\u52a1\u4e0a\u4f9d\u7136\u80dc\u8fc7\u5d4c\u5165\u6a21\u578b，\u5373\u4fbf\u5728 2026 \u5e74\u3002\u7c7b\u578b： \u6784\u5efa
      \n\u8bed\u8a00： Python
      \n\u524d\u7f6e\u6761\u4ef6： Phase 5 \u00b7 01 (\u6587\u672c\u5904\u7406)，Phase 2 \u00b7 02 (\u4ece\u96f6\u5b9e\u73b0\u7ebf\u6027\u56de\u5f52)
      \n\u7528\u65f6： ~75 \u5206\u949f
      \n\u95ee\u9898\u6240\u5728
      \n\u6a21\u578b\u9700\u8981\u6570\u5b57，\u800c\u4f60\u53ea\u6709\u5b57\u7b26\u4e32\u3002
      \n\u6bcf\u4e2a NLP \u7ba1\u9053\u90fd\u5fc5\u987b\u56de\u7b54\u540c\u4e00\u4e2a\u95ee\u9898&#xf<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[81,152,5333,50,224],"topic":[],"class_list":["post-112444","post","type-post","status-publish","format-standard","hentry","category-server","tag-python","tag-pytorch","tag-tf-idf","tag-50","tag-224"],"yoast_head":"\nNLP\u57fa\u7840\u5230\u9ad8\u7ea702\uff1a\u8bcd\u888b\u6a21\u578b\u3001TF-IDF \u4e0e\u6587\u672c\u8868\u793a - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.wsisp.com\/helps\/112444.html\" \/>\n<meta property=\"og:locale\" content=\"zh_CN\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"NLP\u57fa\u7840\u5230\u9ad8\u7ea702\uff1a\u8bcd\u888b\u6a21\u578b\u3001TF-IDF \u4e0e\u6587\u672c\u8868\u793a - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"og:description\" content=\"\u8bcd\u888b\u6a21\u578b\u3001TF-IDF \u4e0e\u6587\u672c\u8868\u793a\u5148\u8ba1\u6570，\u540e\u601d\u8003\u3002TF-IDF \u5728\u5b9a\u4e49\u660e\u786e\u7684\u4efb\u52a1\u4e0a\u4f9d\u7136\u80dc\u8fc7\u5d4c\u5165\u6a21\u578b，\u5373\u4fbf\u5728 2026 \u5e74\u3002\u7c7b\u578b： \u6784\u5efa \u8bed\u8a00： Python \u524d\u7f6e\u6761\u4ef6： Phase 5 \u00b7 01 (\u6587\u672c\u5904\u7406)，Phase 2 \u00b7 02 (\u4ece\u96f6\u5b9e\u73b0\u7ebf\u6027\u56de\u5f52) \u7528\u65f6： ~75 \u5206\u949f \u95ee\u9898\u6240\u5728 \u6a21\u578b\u9700\u8981\u6570\u5b57，\u800c\u4f60\u53ea\u6709\u5b57\u7b26\u4e32\u3002 \u6bcf\u4e2a NLP \u7ba1\u9053\u90fd\u5fc5\u987b\u56de\u7b54\u540c\u4e00\u4e2a\u95ee\u9898&#xf\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.wsisp.com\/helps\/112444.html\" \/>\n<meta property=\"og:site_name\" content=\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"article:published_time\" content=\"2026-10-03T09:16:52+00:00\" \/>\n<meta name=\"author\" content=\"admin\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4\" \/>\n\t<meta name=\"twitter:data2\" content=\"4 \u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/112444.html\",\"url\":\"https:\/\/www.wsisp.com\/helps\/112444.html\",\"name\":\"NLP\u57fa\u7840\u5230\u9ad8\u7ea702\uff1a\u8bcd\u888b\u6a21\u578b\u3001TF-IDF \u4e0e\u6587\u672c\u8868\u793a - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"isPartOf\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\"},\"datePublished\":\"2026-10-03T09:16:52+00:00\",\"dateModified\":\"2026-10-03T09:16:52+00:00\",\"author\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\"},\"breadcrumb\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/112444.html#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.wsisp.com\/helps\/112444.html\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/112444.html#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/www.wsisp.com\/helps\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"NLP\u57fa\u7840\u5230\u9ad8\u7ea702\uff1a\u8bcd\u888b\u6a21\u578b\u3001TF-IDF \u4e0e\u6587\u672c\u8868\u793a\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\",\"url\":\"https:\/\/www.wsisp.com\/helps\/\",\"name\":\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"description\":\"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\",\"name\":\"admin\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"contentUrl\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"caption\":\"admin\"},\"sameAs\":[\"http:\/\/wp.wsisp.com\"],\"url\":\"https:\/\/www.wsisp.com\/helps\/author\/admin\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"NLP\u57fa\u7840\u5230\u9ad8\u7ea702\uff1a\u8bcd\u888b\u6a21\u578b\u3001TF-IDF \u4e0e\u6587\u672c\u8868\u793a - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.wsisp.com\/helps\/112444.html","og_locale":"zh_CN","og_type":"article","og_title":"NLP\u57fa\u7840\u5230\u9ad8\u7ea702\uff1a\u8bcd\u888b\u6a21\u578b\u3001TF-IDF \u4e0e\u6587\u672c\u8868\u793a - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","og_description":"\u8bcd\u888b\u6a21\u578b\u3001TF-IDF \u4e0e\u6587\u672c\u8868\u793a\u5148\u8ba1\u6570，\u540e\u601d\u8003\u3002TF-IDF \u5728\u5b9a\u4e49\u660e\u786e\u7684\u4efb\u52a1\u4e0a\u4f9d\u7136\u80dc\u8fc7\u5d4c\u5165\u6a21\u578b，\u5373\u4fbf\u5728 2026 \u5e74\u3002\u7c7b\u578b： \u6784\u5efa \u8bed\u8a00： Python \u524d\u7f6e\u6761\u4ef6： Phase 5 \u00b7 01 (\u6587\u672c\u5904\u7406)，Phase 2 \u00b7 02 (\u4ece\u96f6\u5b9e\u73b0\u7ebf\u6027\u56de\u5f52) \u7528\u65f6： ~75 \u5206\u949f \u95ee\u9898\u6240\u5728 \u6a21\u578b\u9700\u8981\u6570\u5b57，\u800c\u4f60\u53ea\u6709\u5b57\u7b26\u4e32\u3002 \u6bcf\u4e2a NLP \u7ba1\u9053\u90fd\u5fc5\u987b\u56de\u7b54\u540c\u4e00\u4e2a\u95ee\u9898&#xf","og_url":"https:\/\/www.wsisp.com\/helps\/112444.html","og_site_name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","article_published_time":"2026-10-03T09:16:52+00:00","author":"admin","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005":"admin","\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4":"4 \u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.wsisp.com\/helps\/112444.html","url":"https:\/\/www.wsisp.com\/helps\/112444.html","name":"NLP\u57fa\u7840\u5230\u9ad8\u7ea702\uff1a\u8bcd\u888b\u6a21\u578b\u3001TF-IDF \u4e0e\u6587\u672c\u8868\u793a - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","isPartOf":{"@id":"https:\/\/www.wsisp.com\/helps\/#website"},"datePublished":"2026-10-03T09:16:52+00:00","dateModified":"2026-10-03T09:16:52+00:00","author":{"@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41"},"breadcrumb":{"@id":"https:\/\/www.wsisp.com\/helps\/112444.html#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.wsisp.com\/helps\/112444.html"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.wsisp.com\/helps\/112444.html#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/www.wsisp.com\/helps"},{"@type":"ListItem","position":2,"name":"NLP\u57fa\u7840\u5230\u9ad8\u7ea702\uff1a\u8bcd\u888b\u6a21\u578b\u3001TF-IDF \u4e0e\u6587\u672c\u8868\u793a"}]},{"@type":"WebSite","@id":"https:\/\/www.wsisp.com\/helps\/#website","url":"https:\/\/www.wsisp.com\/helps\/","name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","description":"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41","name":"admin","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/","url":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","contentUrl":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","caption":"admin"},"sameAs":["http:\/\/wp.wsisp.com"],"url":"https:\/\/www.wsisp.com\/helps\/author\/admin"}]}},"_links":{"self":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/112444","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/comments?post=112444"}],"version-history":[{"count":0,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/112444\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media?parent=112444"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/categories?post=112444"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/tags?post=112444"},{"taxonomy":"topic","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/topic?post=112444"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}