{"id":2281,"date":"2021-02-20T10:46:33","date_gmt":"2021-02-20T02:46:33","guid":{"rendered":"http:\/\/www.storagelab.org.cn\/zhangdi\/?p=2281"},"modified":"2021-02-20T10:46:33","modified_gmt":"2021-02-20T02:46:33","slug":"nlp%e5%85%a5%e9%97%a8%e5%b0%8f%e7%bb%93","status":"publish","type":"post","link":"https:\/\/www.divis.cn\/index.php\/2021\/02\/20\/nlp%e5%85%a5%e9%97%a8%e5%b0%8f%e7%bb%93\/","title":{"rendered":"NLP\u5165\u95e8\u5c0f\u7ed3"},"content":{"rendered":"<p>\u7531\u4e8e\u9879\u76ee\u539f\u56e0\u8981\u5e94\u7528\u4e00\u4e9b\u6587\u672c\u5206\u7c7b\u7684\u6280\u672f\uff0c\u4f46\u4e0d\u662f\u7814\u7a76\u8fd9\u4e2a\u65b9\u5411\u3002\u4ece\u7eaf\u5b9e\u7528\u7684\u89d2\u5ea6\uff0c\u67e5\u4e86\u4e00\u4e9b\u8d44\u6599\uff0c\u505a\u4e86\u4ee5\u4e0b\u8bb0\u5f55\u3002<\/p>\n<p>\u673a\u5668\u5b66\u4e60\u7684\u5e93\u73b0\u5728\u6709\u5f88\u591a\uff0c\u4f8b\u5982 Tensorflow\uff0cKeras \u548c Pytorch\u3002\u00a0\u4f46\u662f\uff0c\u5982\u679c\u4e0d\u61c2NLP\u57fa\u7840\uff0c\u4e0d\u77e5\u9053NLP\u5206\u6790\u4efb\u52a1\uff0c\u800c\u76f4\u63a5\u7528\u8fd9\u4e9b\u6846\u67b6\uff0c\u786e\u5b9e\u6709\u4e0d\u77e5\u9053\u5982\u4f55\u4e0a\u624b\u7684\u95ee\u9898\u3002<\/p>\n<p><strong>NLP <\/strong><strong>\u7b80\u4ecb<\/strong><\/p>\n<p>\u81ea\u7136\u8bed\u8a00\u5904\u7406( Natural Language Processing, NLP)\u662f<a href=\"https:\/\/baike.baidu.com\/item\/%E8%AE%A1%E7%AE%97%E6%9C%BA\">\u8ba1\u7b97\u673a<\/a>\u79d1\u5b66\u9886\u57df\u4e0e<a href=\"https:\/\/baike.baidu.com\/item\/%E4%BA%BA%E5%B7%A5%E6%99%BA%E8%83%BD\/9180\">\u4eba\u5de5\u667a\u80fd<\/a>\u9886\u57df\u4e2d\u7684\u4e00\u4e2a\u91cd\u8981\u65b9\u5411\u3002\u5b83\u7814\u7a76\u80fd\u5b9e\u73b0\u4eba\u4e0e<a href=\"https:\/\/baike.baidu.com\/item\/%E8%AE%A1%E7%AE%97%E6%9C%BA\/140338\">\u8ba1\u7b97\u673a<\/a>\u4e4b\u95f4\u7528\u81ea\u7136\u8bed\u8a00\u8fdb\u884c\u6709\u6548\u901a\u4fe1\u7684\u5404\u79cd\u7406\u8bba\u548c\u65b9\u6cd5\u3002\u5176\u4e3b\u8981\u5904\u7406\u7684\u5de5\u4f5c\u5982\u4e0b\uff1a<\/p>\n<p><strong>NLP <\/strong><strong>\u7684\u4efb\u52a1<\/strong><\/p>\n<ol>\n<li><strong>text classification, \u6587\u672c\u5206\u7c7b\uff0c<\/strong>\u5c31\u662f\u6309\u7167\u4e00\u5b9a\u6807\u51c6\u5bf9\u6587\u672c\u96c6\u5408\u8fdb\u884c\u81ea\u52a8\u5206\u7c7b\u6807\u8bb0\u3002\u8fd9\u662fNLP\u5e94\u7528\u6700\u4e3a\u5e7f\u6cdb\u7684\u5e94\u7528\u3002\u73b0\u5728\u901a\u5e38\u89e3\u51b3\u65b9\u6cd5\u662f\u2014\u2014\u4f7f\u7528\u4eba\u5de5\u6807\u6ce8\u8fc7\u7684\u6587\u672c\u96c6\u5408\u4f5c\u4e3a\u8bad\u7ec3\u96c6\uff0c\u8bad\u7ec3\u795e\u7ecf\u7f51\u7edc\u6a21\u578b\uff0c\u7136\u540e\u7528\u8bad\u7ec3\u597d\u7684\u6a21\u578b\u53bb\u81ea\u52a8\u5206\u7c7b\u5176\u4ed6\u6587\u672c\u3002<\/li>\n<li><strong>Sentiment analysis\uff0c\u60c5\u611f\u5206\u6790\uff0c<\/strong>\u5c31\u662f\u5224\u65ad\u6587\u7ae0\u7684\u60c5\u611f\u53d6\u5411\uff08\u79ef\u6781\/\u6d88\u6781\u6216\u8005\u66f4\u591a\u5206\u7c7b\uff09\u3002\u662f\u6587\u672c\u5206\u7c7b\u7684\u4e00\u4e2a\u5206\u652f\u3002<\/li>\n<li><strong>Information Extraction\uff0c\u4fe1\u606f\u63d0\u53d6\uff0c<\/strong>\u662f\u4ece\u975e\u7ed3\u6784\u5316\u6587\u672c\u6e90\u4e2d\u63d0\u53d6\u7279\u5b9a\u4fe1\u606f\u7684\u8fc7\u7a0b\uff0c\u4ee5\u4fbf\u80fd\u591f\u627e\u5230\u5b9e\u4f53\u5e76\u5c06\u5176\u5206\u7c7b\u5e76\u5b58\u50a8\u5728\u6570\u636e\u5e93\u4e2d\u3002\u5178\u578b\u5e94\u7528\u5982\u4ece\u7535\u5b50\u90ae\u4ef6\u4e2d\u63d0\u53d6\u6570\u636e\u6dfb\u52a0\u5230\u65e5\u5386\u91cc\u3002<\/li>\n<li><strong>entity detection, \u5b9e\u4f53\u8bc6\u522b\uff0c<\/strong>\u662f\u4fe1\u606f\u63d0\u53d6\u7684\u4f4e\u7ea7\u5b50\u4efb\u52a1\uff0c\u662f\u4ece\u6587\u672c\u4e2d\u63d0\u53d6\u51fa\u8ba4\u4e3a\u5b9a\u4e49\u7684\u547d\u540d\u5b9e\u4f53\uff0c\u4f8b\u5982\u7ec4\u7ec7\u540d\u79f0\u3001\u673a\u6784\u4ee3\u7801\u3001\u8d27\u5e01\u4ef7\u503c\u3001\u767e\u5206\u6bd4\u3001\u65f6\u95f4\u8868\u8fbe\u7b49\u7b49\u3002<\/li>\n<li><strong>semantic annotation\uff0c\u8bed\u4e49\u6ce8\u91ca\uff0c<\/strong>\u662f\u4fe1\u606f\u63d0\u53d6\u7684\u9ad8\u7ea7\u5b50\u4efb\u52a1\uff0c\u5176\u76ee\u7684\u662f\u5c06\u5b9e\u4f53\u4e0e\u5176\u8bed\u4e49\u63cf\u8ff0\u76f8\u7ed3\u5408\u3002\u4f8b\u5982\uff0c\u5728\u4e00\u7bc7\u6587\u7ae0\u5185\u5230\u5904\u52a0\u5907\u6ce8\uff0c\u5c06\u5185\u5bb9\u94fe\u63a5\u5230\u76f8\u5173\u5b9e\u4f53\u4e0a\uff0c\u4ee5\u4f9b\u8bfb\u8005\u6216\u8005\u5176\u4ed6\u673a\u5668\u5b66\u4e60\u6a21\u578b\u53c2\u8003\u3002\u8fd9\u4e5f\u662f\u77e5\u8bc6\u56fe\u8c31\u7684\u5fc5\u8981\u6280\u672f\u3002<\/li>\n<li><strong>Automatic summarization, \u81ea\u52a8\u6458\u8981\uff0c<\/strong>\u5c31\u50cf\u5c0f\u5b66\u751f\u505a\u9605\u8bfb\u7406\u89e3\u4e00\u6837\uff0c\u81ea\u52a8\u5c06\u4e00\u7bc7\u6587\u7ae0\u7684\u6838\u5fc3\u5185\u5bb9\u603b\u7ed3\u51fa\u6765\u3002\u5c5e\u4e8e\u4fe1\u606f\u63d0\u53d6\u7684\u9ad8\u7ea7\u5e94\u7528\u3002<\/li>\n<li><strong>machine translation, \u673a\u5668\u7ffb\u8bd1\uff0c<\/strong>\u5c31\u662f\u81ea\u52a8\u5c06\u4e00\u79cd\u8bed\u8a00\u7684\u6587\u6863\u7ffb\u8bd1\u6210\u53e6\u5916\u4e00\u79cd\u8bed\u8a00\u3002\u5176\u4e3b\u8981\u96be\u70b9\u5728\u4e8e\u4eba\u7c7b\u8bed\u8a00\u5177\u6709\u6b67\u4e49\u6027\uff0c\u56e0\u6b64\u597d\u7684\u7ffb\u8bd1\u4e0d\u4ec5\u4ec5\u5728\u4e8e\u53c2\u8003\u4e0a\u4e0b\u6587\uff0c\u800c\u662f\u8981\u53c2\u8003\u5de8\u91cf\u7684\u4eba\u7c7b\u8bed\u6599\u96c6\u3002<\/li>\n<li><strong>Question answering systems\uff0c \u81ea\u52a8\u56de\u7b54\u7cfb\u7edf\uff0c<\/strong>\u6839\u636e\u7528\u6237\u7684\u63d0\u95ee\u8fdb\u884c\u81ea\u52a8\u56de\u7b54\u3002\u8fd9\u662f\u6700\u80fd\u4f53\u73b0AI\u6027\u80fd\u7684\u5e94\u7528\uff08\u56fe\u7075\u6d4b\u8bd5\uff09\u3002\u5178\u578b\u5e94\u7528\u5982\u9762\u8bd5\u673a\u5668\u4eba\u3001\u5fae\u4fe1\u81ea\u52a8\u56de\u7b54\u5c0f\u52a9\u624b\u7b49\u7b49\u3002\u663e\u7136\u5b83\u6bd4\u673a\u5668\u7ffb\u8bd1\u8fd8\u96be\uff0c\u4e0d\u4ec5\u8981\u7406\u89e3\u7528\u6237\u7684\u95ee\u9898\uff0c\u8fd8\u5f97\u7ec4\u7ec7\u56de\u7b54\u3002<\/li>\n<li>\u7b49\u7b49<\/li>\n<\/ol>\n<p><strong>NLP<\/strong><strong>\u4e2d\u6587\u672c\u5206\u7c7b\u4efb\u52a1\u7684\u4e00\u822c\u5904\u7406\u6d41\u7a0b<\/strong><\/p>\n<p>\u6211\u4eec\u4ee5\u8981\u505a\u7684\u6587\u672c\u5206\u7c7b\u4efb\u52a1\u4e3a\u4f8b\uff0c\u4ecb\u7ecd\u4e00\u4e0b\u5e38\u89c1\u7684\u57fa\u4e8e\u673a\u5668\u5b66\u4e60\u7684\u6587\u672c\u5206\u7c7b\u5904\u7406\u6d41\u7a0b\uff1a<\/p>\n<p><img decoding=\"async\" src=\"http:\/\/zhangdi.dengxinbo.cn\/NLP.jpg\" \/><\/p>\n<ol>\n<li><strong>\u51c6\u5907\u6570\u636e\uff1a<\/strong>\u4e3a\u4e86\u5f97\u5230\u673a\u5668\u5b66\u4e60\u7684\u6700\u7ec8\u6210\u679c\u2014\u2014\u53ef\u4ee5\u81ea\u52a8\u8fdb\u884c\u6587\u672c\u5206\u7c7b\u7684\u5206\u7c7b\u5668\uff0c\u9996\u5148\u6211\u4eec\u8981\u51c6\u5907\u8bad\u7ec3\u6570\u636e\uff0c\u5373\u8bad\u7ec3\u96c6\u3002\u4e00\u822c\u6765\u8bf4\u7528\u4f5c\u8bad\u7ec3\u96c6\u7684\u6570\u636e\uff0c\u9700\u8981\u662f\u5df2\u7ecf\u4eba\u5de5\u6807\u6ce8\u7684\u3002\u673a\u5668\u5b66\u4e60\u6a21\u578b\u901a\u8fc7\u8fd9\u4e9b\u6570\u636e\u7684\u8bad\u7ec3\uff0c\u624d\u80fd\u77e5\u9053\u600e\u4e48\u5206\u7c7b\u662f\u5bf9\uff0c\u600e\u4e48\u5206\u7c7b\u662f\u9519\uff0c\u8fdb\u800c\u505a\u5230\u673a\u5668\u81ea\u52a8\u5206\u7c7b\u3002\u5728AI\u754c\u6709\u4e00\u53e5\u8bdd\uff0c&#8221;\u6570\u636e\u51b3\u5b9a\u4e86\u4f60\u7684\u6a21\u578b\u4e0a\u9650\uff0c\u800c\u6a21\u578b\u53ea\u662f\u5728\u903c\u8fd1\u8fd9\u4e2a\u4e0a\u9650&#8221;\u3002\u53ef\u89c1\u6570\u636e\u7684\u91cd\u8981\u3002\u7136\u800c\u4eba\u5de5\u6807\u6ce8\u6570\u636e\u662f\u4e00\u4ef6\u5341\u5206\u8017\u8d39\u4eba\u529b\u7684\u4e8b\uff0c\u6240\u4ee5\u6709\u6709\u5f88\u591a\u6280\u672f\u548c\u65b9\u6cd5\u6765\u6539\u8fdb\u8fd9\u4e2a\u95ee\u9898\uff08\u4f8b\u5982\u4f7f\u7528\u53ef\u89c6\u5316\u6280\u672f\u52a0\u5feb\u6807\u6ce8\u6548\u7387\u3001\u4f7f\u7528\u673a\u5668\u5b66\u4e60\u6a21\u578b\u751f\u6210\u6807\u6ce8\u597d\u7684\u6570\u636e\u7b49\u7b49\uff09\u3002\u8fd8\u597d\uff0c\u5165\u95e8NLP\u5e76\u4e0d\u9700\u8981\u6211\u4eec\u4eba\u5de5\u6807\u6ce8\u6570\u636e\uff0c\u5df2\u7ecf\u6709\u5f88\u591a\u5df2\u7ecf\u6807\u6ce8\u597d\u7684\u8bad\u7ec3\u96c6\u53ef\u4f9b\u4e0b\u8f7d\u3002\u8fd9\u6837\u7684\u8bad\u7ec3\u96c6\u7279\u522b\u591a\uff0c\u968f\u4fbf\u5728\u77e5\u4e4e\u4e0a\u641c\u4e00\u4e0b\u5c31\u6709\uff08<a href=\"https:\/\/www.zhihu.com\/search?type=content&amp;q=nlp%20%E6%95%B0%E6%8D%AE%E9%9B%86\">https:\/\/www.zhihu.com\/search?type=content&amp;q=nlp%20%E6%95%B0%E6%8D%AE%E9%9B%86<\/a>\uff09\u3002\u7531\u4e8e\u5206\u7684\u5f88\u7ec6\uff0c\u6240\u4ee5\u4f7f\u7528\u65f6\u6ce8\u610f\u662f\u5426\u4e0e\u4f60\u7684\u5206\u6790\u4efb\u52a1\u76f4\u63a5\u76f8\u5173\u3002\u5f53\u4f60\u719f\u6089\u4e86\u6574\u4e2a\u6587\u672c\u5206\u7c7b\u4e86\u6d41\u7a0b\u540e\uff0c\u4e3a\u4e86\u89e3\u51b3\u5b9e\u9645\u95ee\u9898\uff0c\u53ef\u80fd\u8fd8\u9700\u8981\u66f4\u591a\u7684\u6807\u6ce8\u6570\u636e\uff0c\u8fd9\u4e2a\u65f6\u5019\u5c31\u9700\u8981\u4f60\u51c6\u5907\u81ea\u5df1\u7684\u4eba\u5de5\u6807\u6ce8\u6570\u636e\u4e86\u3002<\/li>\n<li><strong>\u6570\u636e\u9884\u5904\u7406\uff1a<\/strong>\u51c6\u5907\u597d\u6570\u636e\u540e\uff0c\u63a5\u4e0b\u6765\u5c31\u662f\u6570\u636e\u9884\u5904\u7406\u3002\u7f51\u4e0a\u7684\u5f00\u6e90\u8bad\u7ec3\u96c6\uff0c\u6709\u7684\u5df2\u7ecf\u7ecf\u8fc7\u4e86\u9884\u5904\u7406\uff08\u4f8b\u5982\u5df2\u7ecf\u5206\u8bcd\u3001\u53bb\u9664\u505c\u7528\u8bcd\u7b49\u7b49\uff09\uff0c\u4f46\u6709\u7684\u6709\u6ca1\u6709\u3002\u5e76\u4e14\u56e0\u4e3a\u5206\u7c7b\u4efb\u52a1\u7684\u4e0d\u540c\uff0c\u53ef\u80fd\u9700\u8981\u7684\u9884\u5904\u7406\u6b65\u9aa4\u4f1a\u6709\u533a\u522b\u3002\u4f8b\u5982\u8fdb\u884c\u60c5\u611f\u5206\u6790\uff0c\u53ef\u80fd\u8fd8\u8981\u6784\u9020\u60c5\u611f\u8bcd\u5178\u3002\u4e3a\u4e86\u8fdb\u884c\u81ea\u52a8\u6458\u8981\uff0c\u53ef\u80fd\u9700\u8981\u5148\u505a\u5b9e\u4f53\u68c0\u6d4b\u3002\u6240\u4ee5\u6570\u636e\u9884\u5904\u7406\u6b65\u9aa4\u4e0d\u80fd\u7701\u7565\u3002\u5e7f\u4e49\u7684\u6570\u636e\u9884\u5904\u7406\u8fd8\u5305\u62ec\u7279\u5f81\u63d0\u53d6\u8fd9\u4e00\u6b65\u9aa4\uff0c\u4f46\u7531\u4e8e\u7279\u5f81\u63d0\u53d6\u8f83\u4e3a\u91cd\u8981\uff0c\u6240\u4ee5\u4e0b\u6587\u5355\u72ec\u8bb2\u4e00\u4e0b\u3002\u4e3a\u4e86\u7b80\u4fbf\u8d77\u89c1\uff0c\u9884\u5904\u7406\u901a\u5e38\u4f1a\u4f7f\u7528\u4e00\u4e9b\u4e13\u95e8\u7684\u7c7b\u5e93\u6765\u505a\uff0c\u4f8b\u5982spaCy\uff0cTorchText\u7b49\u3002\u8fd9\u4e9b\u5e93\u53ef\u4ee5\u4ece\u5206\u8bcd\u4e00\u76f4\u5904\u7406\u5230\u8bcd\u5411\u91cf\uff0c\u7136\u540e\u4ea4\u7ed9\u673a\u5668\u5b66\u4e60\u6a21\u578b\u53bb\u5b66\u4e60\u3002<\/li>\n<li><strong>\u7279\u5f81\u63d0\u53d6\uff1a<\/strong>\u7279\u5f81\u63d0\u53d6\u5c31\u662f\u628a\u5904\u7406\u597d\u7684\u6570\u636e\uff0c\u7528\u673a\u5668\u53ef\u4ee5\u770b\u61c2\u7684\u6700\u5c0f\u7ed3\u6784\u8868\u8fbe\u51fa\u6765\uff0c\u4ece\u800c\u65b9\u4fbf\u540e\u7eed\u7684\u8ba1\u7b97\u3002\u5728NLP\u6587\u672c\u5206\u7c7b\u4efb\u52a1\u4e2d\uff0c\u5e38\u89c1\u7684\u7279\u5f81\u63d0\u53d6\u65b9\u5f0f\u6709\u4e24\u5927\u7c7b\uff1a\u8bcd\u888b\u6a21\u578b\u548c\u8bcd\u5411\u91cf\u6a21\u578b\u3002\u8bcd\u888b\u6a21\u578b\u4e2d\u5e38\u7528\u7684\u65b9\u6848\u662fTF-IDF\u6a21\u578b\uff0c\u8bcd\u5411\u91cf\u6a21\u578b\u5e38\u7528\u7684\u65b9\u6848\u6709one-hot\u548cword2vec\u3002\u6211\u4eec\u4f5c\u4e3a\u4f7f\u7528\u8005\u53ea\u8981\u767e\u5ea6\u8c37\u6b4c\u4ee5\u4e0b\u77e5\u9053\u5176\u6982\u5ff5\u5373\u53ef\uff0c\u5e76\u4e14\u4e86\u89e3\u6211\u4eec\u7684\u6d4b\u8bd5\u6848\u4f8b\u7528\u7684\u662f\u54ea\u4e00\u4e2a\u5c31\u884c\u3002\u7ecf\u8fc7\u7279\u5f81\u63d0\u53d6\uff0c\u8bad\u7ec3\u96c6\u6570\u636e\u5c06\u8f6c\u6362\u4e3aTF-IDF\u7684\u8bcd\u888b\u6570\u636e\u6216\u8005word2vec\u7684\u9ad8\u7ef4\u8bcd\u5411\u91cf\uff0c\u8fd9\u6837\u6587\u672c\u5206\u7c7b\u7684\u673a\u5668\u5b66\u4e60\u6a21\u578b\u624d\u597d\u8fdb\u884c\u5904\u7406\u3002<\/li>\n<li><strong>\u6a21\u578b\u8bad\u7ec3\uff1a<\/strong>\u7528\u4e8e\u6587\u672c\u5206\u6790\u7684\u673a\u5668\u5b66\u4e60\u6a21\u578b\u6709\u5f88\u591a\uff0c\u4f8b\u5982KNN\uff0cSVN\uff0cK-mean\uff0cCNN\uff0cLSTM\u7b49\u7b49\u3002\u4f7f\u7528 Tensorflow\uff0cKeras \u548c Pytorch\u8fd9\u6837\u7684\u673a\u5668\u5b66\u4e60\u6846\u67b6\uff0c\u53ef\u4ee5\u8f83\u4e3a\u5bb9\u6613\u5730\u8c03\u7528\u8fd9\u4e9b\u6a21\u578b\uff0c\u53bb\u5b66\u4e60\u4f60\u5904\u7406\u597d\u7684\u8bcd\u888b\u6216\u8005\u8bcd\u5411\u91cf\uff0c\u8fd9\u4e2a\u8fc7\u7a0b\u5c31\u662f\u6a21\u578b\u8bad\u7ec3\u3002\u6839\u636e\u4f60\u8bad\u7ec3\u96c6\u7684\u89c4\u6a21\u548c\u673a\u5668\u7684\u6027\u80fd\u4e0d\u540c\uff0c\u6a21\u578b\u5b8c\u6574\u8bad\u7ec3\u4e00\u6b21\u7684\u65f6\u95f4\u4ece\u51e0\u5206\u949f\u5230\u51e0\u767e\u5c0f\u65f6\u4e0d\u7b49\uff0c\u6240\u4ee5\u6700\u597d\u91cf\u529b\u800c\u884c\u3002\u6b64\u5916\u521d\u671f\u5b66\u4e60\u65f6\u4e0d\u8981\u7ea0\u7ed3\u8fd9\u4e9b\u673a\u5668\u5b66\u4e60\u6a21\u578b\u7684\u539f\u7406\uff0c\u5c31\u628a\u5b83\u5f53\u4f5c\u4e00\u4e2a\u7535\u89c6\u673a\uff0c\u641e\u6e05\u695a\u600e\u4e48\u64cd\u4f5c\u9065\u63a7\u5668\u5c31\u53ef\u4ee5\u4e86\u3002<\/li>\n<li><strong>\u6a21\u578b\u9884\u6d4b\u7ed3\u679c\u8bc4\u4ef7\uff1a<\/strong>\u6a21\u578b\u7ecf\u8fc7\u8bad\u7ec3\uff0c\u5c31\u53ef\u4ee5\u81ea\u5df1\u53bb\u6587\u672c\u5206\u7c7b\u4e86\u3002\u6211\u4eec\u5f53\u7136\u5e0c\u671b\u6a21\u578b\u4e0d\u4ec5\u4ec5\u662f\u5728\u8bad\u7ec3\u96c6\u4e0a\u624d\u4f1a\u9884\u6d4b\uff0c\u5bf9\u4e8e\u975e\u8bad\u7ec3\u96c6\u7684\u76ee\u6807\u6570\u636e\uff08\u79f0\u4e3a\u6d4b\u8bd5\u96c6\uff09\u4e5f\u80fd\u6709\u7528\uff0c\u5373\u8fbe\u5230\u6240\u8c13\u4e3e\u4e00\u53cd\u4e09\u7684\u6548\u679c\u3002\u4f46\u662f\u673a\u5668\u5b66\u4e60\u6a21\u578b\u7684\u6587\u672c\u5206\u7c7b\u7ed3\u679c\uff0c\u8ddf\u6807\u51c6\u7b54\u6848\u53ef\u80fd\u4e0d\u5b8c\u5168\u4e00\u81f4\uff0c\u751a\u81f3\u53ef\u80fd\u9519\u5f97\u79bb\u8c31\u3002\u9700\u8981\u6709\u5bf9\u6a21\u578b\u7684\u8bc4\u6d4b\u6307\u6807\uff0c\u6700\u91cd\u8981\u7684\u8bc4\u6d4b\u6307\u6807\u5c31\u662f\u51c6\u786e\u7387\uff08\u8ba1\u7b97\u6a21\u578b\u6b63\u786e\u5206\u7c7b\u7684\u6837\u672c\u6570\u4e0e\u603b\u6837\u672c\u6570\u4e4b\u6bd4\u4ee5\u8861\u91cf\u6a21\u578b\u7684\u6548\u679c\uff09\u3002 \u5176\u4ed6\u8fd8\u6709\u7cbe\u786e\u7387\uff0c\u53ec\u56de\u7387\uff0c F1\u7b49\u7b49\u3002<\/li>\n<li><strong>\u6b20\u62df\u5408\u4e0e\u8fc7\u62df\u5408\uff1a<\/strong>\u5bf9\u4e8e\u5de5\u4e1a\u5316\u5e94\u7528\uff0c\u5f53\u7136\u662f\u51c6\u786e\u7387\u8d8a\u9ad8\u8d8a\u597d\u3002\u4e0d\u8fc7\uff0c\u6a21\u578b\u5bf9\u4e8e\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\u7684\u51c6\u786e\u7387\u901a\u5e38\u4e0d\u4e00\u6837\uff0c\u4ee5\u4e0b\u4e24\u79cd\u60c5\u51b5\u662f\u8981\u7740\u529b\u907f\u514d\u7684\uff1a\u5176\u4e00\uff0c\u6a21\u578b\u5728\u8bad\u7ec3\u96c6\u4e0a\u7684\u51c6\u786e\u7387\u5f88\u4f4e\uff0c\u79f0\u4e4b\u4e3a\u6b20\u62df\u5408\u3002\u8bf4\u660e\u8be5\u6a21\u578b\u4e0d\u662f\u5f88\u9002\u5408\u5b66\u4e60\u8fd9\u4e2a\u8bad\u7ec3\u96c6\u7684\u6570\u636e\u3002\u89e3\u51b3\u65b9\u6848\u662f\u6362\u6a21\u578b\u3001\u589e\u52a0\u6a21\u578b\u590d\u6742\u5ea6\u3001\u51cf\u5c11\u6b63\u5219\u5316\u53c2\u6570\u7b49\u3002\u5176\u4e8c\uff0c\u6a21\u578b\u5728\u8bad\u7ec3\u96c6\u4e0a\u51c6\u786e\u7387\u5f88\u9ad8\uff0c\u4f46\u5728\u6d4b\u8bd5\u96c6\u4e0a\u7684\u51c6\u786e\u7387\u5f88\u4f4e\uff0c\u79f0\u4e4b\u4e3a\u8fc7\u62df\u5408\u3002\u8fd9\u79cd\u60c5\u51b5\u8f83\u4e3a\u5e38\u89c1\uff0c\u8bf4\u660e\u6a21\u578b\u53ea\u5bf9\u7f3a\u4e4f\u4e3e\u4e00\u53cd\u4e09\u7684\u80fd\u529b\u3002\u89e3\u51b3\u65b9\u6848\u6709\u589e\u5927\u8bad\u7ec3\u96c6\u7684\u6570\u636e\u91cf\u3001\u589e\u52a0\u6b63\u5219\u5316\u53c2\u6570\u3001\u4eba\u5de5\u6539\u5584\u7279\u5f81\u9009\u53d6\u7b49\u3002<\/li>\n<li><strong>\u5206\u7c7b\u5668\u4e0a\u7ebf\u5e94\u7528\uff1a<\/strong>\u673a\u5668\u5b66\u4e60\u6a21\u578b\u7ecf\u8fc7\u53cd\u590d\u5730\u8bad\u7ec3\u3001\u8bc4\u4ef7\u548c\u8c03\u6574\uff0c\u6700\u7ec8\u8fbe\u5230\u4e00\u4e2a\u5728\u6d4b\u8bd5\u96c6\u4e0a\u51c6\u786e\u7387\u53ef\u4ee5\u63a5\u53d7\u7684\u72b6\u6001\u3002\u8fd9\u4e2a\u65f6\u5019\u5c31\u53ef\u4ee5\u628a\u6a21\u578b\u505a\u7ebf\u4e0a\u90e8\u7f72\uff0c\u7136\u540e\u8ba9\u4ed6\u670d\u52a1\u4e8e\u4eba\u7c7b\u4e86\u3002\u5f53\u7136\u5bf9\u4e8e\u767e\u5ea6\u8c37\u6b4c\u5f88\u591a\u5927\u516c\u53f8\u6765\u8bf4\uff0c\u8fd8\u4f1a\u505a\u5728\u7ebf\u8bad\u7ec3\uff0c\u5373\u5728\u6a21\u578b\u90e8\u7f72\u540e\uff0c\u8fd8\u4f1a\u6301\u7eed\u5730\u4ee5\u4eba\u4eec\u8f93\u5165\u7684\u6570\u636e\u4f5c\u4e3a\u517b\u6599\u8fdb\u884c\u8bad\u7ec3\uff0c\u4e0d\u65ad\u6539\u8fdb\u3002\u6700\u5178\u578b\u7684\u4f8b\u5b50\u5c31\u662f\u673a\u5668\u7ffb\u8bd1\u5728\u6700\u8fd110\u5e74\u4e2d\u4e0d\u65ad\u8fdb\u6b65\uff0c\u4ee5\u81f3\u4e8e\u8ba9\u5f88\u591a\u82f1\u8bed\u4e13\u4e1a\u7684\u4e2d\u4f4e\u7aef\u4eba\u624d\u8f6c\u884c\u3002<\/li>\n<\/ol>\n<p>\u7efc\u4e0a\u53ef\u89c1\uff0c\u5982\u679c\u4e0d\u662f\u7814\u7a76\u673a\u5668\u5b66\u4e60\u6a21\u578b\u800c\u53ea\u662f\u4e3a\u4e86\u5b66\u4f1a\u4f7f\u7528\uff0c\u90a3\u4e48\u6574\u4e2aNLP\u5206\u7c7b\u6587\u672c\u4efb\u52a1\u7684\u5de5\u4f5c\u91cf\u4e3b\u8981\u5728\u6570\u636e\u9884\u5904\u7406\u548c\u6a21\u578b\u8bad\u7ec3\u8c03\u53c2\u8fd9\u4e24\u6b65\u3002<\/p>\n<p><strong>\u6587\u672c\u5206\u7c7b\u76f8\u5173\u9879\u76ee\u63a8\u8350<\/strong><\/p>\n<p>\u767e\u5ea6 Senta \u60c5\u611f\u5206\u6790\u7cfb\u7edf\uff0c\u57fa\u4e8e\u767e\u5ea6AI\u5e73\u53f0\uff0c\u90e8\u7f72\u4f7f\u7528\u7b80\u5355\u3002\u53ef\u4ee5\u5148\u7528\u6765\u7ec3\u624b\u3002\u00a0<a href=\"https:\/\/github.com\/baidu\/Senta\">https:\/\/github.com\/baidu\/Senta<\/a><\/p>\n<p><span style=\"color: #000000\"><b>\u5173\u4e8e\u6570\u636e\u9884\u5904\u7406<\/b><\/span><\/p>\n<p>\u6bcf\u4e00\u7c7bNPL\u4efb\u52a1\uff0c\u7528\u5230\u7684\u5177\u4f53\u65b9\u6cd5\u90fd\u4e0d\u4e00\u6837\uff0c\u4f46\u90fd\u6709\u5171\u6027\u3002\u6700\u5927\u7684\u5171\u6027\u662f\uff0c\u4f60\u90fd\u9700\u8981\u5bf9\u8981\u5904\u7406\u7684\u6587\u672c\u8fdb\u884c\u4e00\u4e9b\u9884\u5904\u7406\uff08\u5206\u53e5\uff0c\u5206\u8bcd\uff0c\u8bc6\u522b\u505c\u7528\u8bcd\uff0c\u5b9e\u4f53\u68c0\u6d4b\uff0c\u4f9d\u5b58\u5206\u6790\u7b49\u7b49\uff09\uff0c\u624d\u80fd\u8fdb\u884c\u9ad8\u7ea7\u64cd\u4f5c\uff08\u6587\u672c\u5206\u7c7b\u3001\u77e5\u8bc6\u63d0\u53d6\u7b49\u7b49\uff09\u3002<\/p>\n<p>\u63a8\u8350\u4e00\u4e2a\u5e93spaCy\uff0c\u662f\u6700\u53d7\u6b22\u8fce\u7684\u6587\u672c\u9884\u5904\u7406\u5e93\u3002\u5b83\u5305\u542b\u8bb8\u591a\u6613\u4e8e\u4f7f\u7528\u7684\u51fd\u6570\uff0c\u53ef\u4ee5\u7528\u4e8e\u6807\u8bb0\u5316\u3001\u8bcd\u6027\u6807\u8bb0\u3001\u547d\u540d\u5b9e\u4f53\u8bc6\u522b\u7b49\u3002\u5b83\u8fd8\u652f\u630159\u79cd\u4ee5\u4e0a\u7684\u8bed\u8a00\u548c\u51e0\u79cd\u9884\u8bad\u7ec3\u7684\u5355\u8bcd\u5411\u91cf\uff0c\u4f60\u5f88\u5bb9\u6613\u5c31\u53ef\u4ee5\u5feb\u901f\u5165\u95e8\uff01\u5730\u5740\uff1a<a href=\"https:\/\/github.com\/explosion\/spaCy\">https:\/\/github.com\/explosion\/spaCy<\/a><\/p>\n<p><img decoding=\"async\" src=\"http:\/\/zhangdi.dengxinbo.cn\/spacy.png\" \/><\/p>\n<p>&nbsp;<\/p>\n<p><strong>SpaCy<\/strong><strong>\u7684\u529f\u80fd\uff1a<\/strong><\/p>\n<ol>\n<li><strong>Sentencizer\uff0c\u65ad\u53e5\uff0c<\/strong>\u5c31\u662f\u5c06\u6587\u7ae0\u5206\u6210\u53e5\u5b50\u3002\u867d\u7136\u73b0\u4ee3\u4e0d\u8bba\u4e2d\u6587\u6587\u7ae0\u8fd8\u662f\u82f1\u6587\u6587\u7ae0\u9760\u53e5\u53f7\u5c31\u80fd\u65ad\u53e5\uff0c\u4f46\u662f\u53ef\u80fd\u5b58\u5728\u4e00\u4e9b\u4e0d\u6613\u5904\u7406\u7684\u7279\u6b8a\u89c4\u5219\u3002\u4f8b\u5982\uff0c\u5f53\u4e00\u7bc7\u82f1\u6587\u6587\u7ae0\u6709\u5f88\u591a\u516c\u5f0f\u7684\u65f6\u5019\uff0c\u4e0d\u597d\u7b80\u5355\u9760\u53e5\u53f7\u533a\u5206\uff0c\u5c31\u53ef\u4ee5\u5b9a\u5236\u5176\u53e5\u5f0f\u7684\u5f00\u5934\u7ed3\u5c3e\u3002\u6b64\u5916\uff0cspaCy\u652f\u6301\u5728\u4ee3\u7801\u91cc\u52a0\u8f7d\u5176\u4ed6\u8bed\u8a00\u6a21\u578b\u5bf9\u6587\u672c\u8fdb\u884c\u5206\u6790\u3002<\/li>\n<li><strong>Tokenization\uff0c\u5206\u8bcd\uff0c<\/strong>\u5c31\u662f\u5c06\u53e5\u5b50\u5206\u6210\u8bcd\u3002\u82f1\u6587\u6bd4\u8f83\u7b80\u5355\u76f4\u63a5\u9760\u7a7a\u683c\u5206\u8bcd\uff0c\u800c\u4e2d\u6587\u8981\u6709\u7279\u6b8a\u7684\u5206\u8bcd\u64cd\u4f5c\u3002spaCy 2.2.3\u4ee5\u540e\u7248\u672c\u652f\u6301\u4e2d\u6587\u3002<\/li>\n<li><strong>Part-of-speech tagging\uff0c\u8bcd\u6027\u6807\u6ce8<\/strong>\uff0c\u7b80\u79f0 POS\u3002\u5c31\u662f\u6807\u6ce8\u5355\u8bcd\u662f\u4ec0\u4e48\u7c7b\u578b\uff0c\u7c7b\u578b\u5305\u62ec\u540d\u8bcd\/\u52a8\u8bcd\uff0c\u4e5f\u53ef\u4ee5\u81ea\u5b9a\u4e49\uff0c\u6bd4\u5982\u566a\u58f0\u8bcd\/\u6709\u7528\u8bcd\u7b49\u7b49\u3002<\/li>\n<li><strong>Lemmatization\uff0c\u8bcd\u6027\u8fd8\u539f\u3002<\/strong>\u5bf9\u4e8e\u82f1\u6587\u3001\u62c9\u4e01\u6587\u7b49\u8bed\u8a00\uff0c\u8bcd\u6c47\u90fd\u6709\u53d8\u683c\u7684\u5f62\u5f0f\uff0c\u8bcd\u6027\u8fd8\u539f\u53ef\u4ee5\u5c06\u53d8\u683c\u540e\u7684\u5355\u8bcd\u8fd8\u539f\u6210\u539f\u5f62\u6001\u4ee5\u4fbf\u5206\u6790\u3002\u4f8b\u5982\uff0c\u590d\u6570\u8fd8\u539f\u6210\u5355\u6570(cats -&gt; cat)\uff0c\u8fc7\u53bb\u65f6\u6001\u8fd8\u539f\u6210\u73b0\u5728\u65f6\u6001 (had -&gt; have)\u3002\u4e2d\u6587\u4e0d\u9700\u8981\u3002<\/li>\n<li><strong>Stop words\uff0c\u8bc6\u522b\u505c\u7528\u8bcd\u3002<\/strong>\u505c\u7528\u8bcd\u4e00\u822c\u662f\u82f1\u6587\u4e2d\u51fa\u73b0\u9891\u7387\u6700\u9ad8\u3001\u4f46\u53c8\u6ca1\u4ec0\u4e48\u8bed\u4e49\u7684\u8bcd\uff0c\u5982 a, the\u3002\u5728\u5f88\u591aNLP\u4efb\u52a1\u5f00\u5c55\u524d\u4e2d\u90fd\u8981\u53bb\u6389\u8fd9\u4e9b\u8bcd\u7684\u5e72\u6270\uff0c\u907f\u514d\u5b83\u4eec\u6d6a\u8d39\u5185\u5b58\u3002<\/li>\n<li><strong>Dependency Parsing\uff0c\u4f9d\u5b58\u5206\u6790\uff0c<\/strong>\u5c31\u662f\u901a\u8fc7\u8bed\u6cd5\u89e3\u6790\u6811\u7684\u6784\u5efa\uff0c\u6807\u8bb0\u5355\u8bcd\u662f\u4e3b\u8bed\uff0c\u8c13\u8bed\uff0c\u5bbe\u8bed\u8fd8\u662f\u8fde\u63a5\u8bcd\u3002\u4ece\u800c\u65b9\u4fbf\u68c0\u6d4b\u53e5\u5b50\u7684\u8fb9\u754c\uff0c\u5207\u5206\u77ed\u8bed\u7b49\u5de5\u4f5c\u3002Spacy \u7684\u4f9d\u5b58\u5206\u6790\u91c7\u7528\u4e86 ClearNLP \u7684\u4f9d\u5b58\u5206\u6790\u6807\u7b7e<a href=\"https:\/\/link.zhihu.com\/?target=https%3A\/\/github.com\/clir\/clearnlp-guidelines\/blob\/master\/md\/specifications\/dependency_labels.md\">ClearNLP Dependency Labels<\/a>\u3002\u6839\u636e\u8fd9\u4e2a\u7f51\u7ad9\u63d0\u4f9b\u7684\u6807\u7b7e\u8bcd\u5178\uff0c\u4e00\u53e5\u8bdd\u53ef\u4ee5\u88ab\u6784\u5efa\u6210\u5982\u4e0b\u5185\u5bb9 [\u9650\u5b9a\u8bcd\uff0c \u5f62\u5bb9\u8bcd\u4fee\u9970, \u540d\u8bcd\u4e3b\u8bed\uff0c \u6839\u8282\u70b9, \u9650\u5b9a\u8bcd, \u5f62\u5bb9\u8bcd\u4fee\u9970, \u5f62\u5bb9\u8bcd\u4fee\u9970, \u5c5e\u6027, \u6807\u70b9] \uff0c\u5176\u4e2d\u6839\u8282\u70b9\u662f\u4e00\u53e5\u8bdd\u7684\u8c13\u8bcd\u3002\u4f9d\u5b58\u5206\u6790\u662f\u8bed\u4e49\u5206\u6790\u7684\u57fa\u7840\uff0c\u975e\u5e38\u91cd\u8981\u3002\u6bd4\u5982\u5728\u79d1\u6280\u8bba\u6587\u5199\u4f5c\u4e2d\u6709\u4e00\u6761\u89c4\u5219\u662f\uff0c\u4e3b\u8bed\u4e0d\u80fd\u8fc7\u957f\uff0c\u8c13\u8bed\uff08\u6839\u8282\u70b9\uff09\u51fa\u73b0\u7684\u4f4d\u7f6e\u4e0d\u80fd\u8fc7\u4e8e\u9760\u540e\uff0c\u5426\u5219\u4f1a\u5f71\u54cd\u9605\u8bfb\u7406\u89e3\u3002\u5177\u4f53\u5b9e\u73b0\u65b9\u6cd5\u5c31\u662f\u5224\u65ad\u6839\u8282\u70b9\u7684\u4f4d\u7f6e\uff0c\u5373\u4e00\u53e5\u8bdd\u4e2d\u52a8\u8bcd\u8c13\u8bed\u7684\u4f4d\u7f6e\uff0c\u5982\u679c\u8fc7\u4e8e\u9760\u540e\uff0c\u5219\u63d0\u51fa\u4fee\u6539\u610f\u89c1\u3002<\/li>\n<li><strong>Named Entity Recognization\uff0c\u5b9e\u4f53\u68c0\u6d4b\uff0c<\/strong>\u7b80\u79f0NER\uff0c\u5c31\u662f\u8bc6\u522b\u51fa\u6587\u6863\u4e2d\u7684\u5b9e\u4f53\u77ed\u8bed\uff0c\u6709\u591a\u79cd\u7c7b\u578b\u7684\u5b9e\u4f53\uff0c\u4f8b\u5982 &#8211; \u4eba\u7269\uff0c\u5730\u70b9\uff0c\u7ec4\u7ec7\uff0c\u65e5\u671f\uff0c\u6570\u5b57\u3002\u4e3a\u4e0b\u4e00\u6b65\u5de5\u4f5c\u505a\u51c6\u5907\u3002SpaCy\u6709\u4e13\u95e8\u7684\u8bcd\u5178\u6765\u505a\u6b64\u4e8b\u3002<\/li>\n<li><strong>Noun Chunks\uff0c\u63d0\u53d6\u540d\u8bcd\u77ed\u8bed\uff0c<\/strong> \u5c31\u662f\u628a\u4e00\u4e9b\u7279\u5b9a\u7684\u540d\u8bcd\u77ed\u8bed\u63d0\u53d6\u51fa\u6765\u3002\u4f8b\u5982\u6587\u7ae0\u4e2d\u7ecf\u5e38\u51fa\u73b0\u201c\u5170\u5dde\u7406\u5de5\u5927\u5b66\u201d\uff0c\u90a3\u4e48\u8fd9\u4e2a\u201c\u5170\u5dde\u7406\u5de5\u5927\u5b66\u201d\u5c31\u662f\u4e00\u4e2a\u7279\u5b9a\u7684\u540d\u8bcd\u77ed\u8bed\u3002<\/li>\n<li><strong>Coreference Resolution\uff0c\u6307\u4ee3\u6d88\u89e3\uff0c<\/strong>\u5bfb\u627e\u53e5\u5b50\u4e2d\u4ee3\u8bcd he\uff0cshe\uff0cit \u6240\u5bf9\u5e94\u7684\u5b9e\u4f53\u3002\u663e\u7136\u60f3\u505a\u5230\u8fd9\u4e00\u70b9\uff0c\u5c31\u4e0d\u80fd\u7528\u89c4\u5219\u3001\u8bcd\u5178\u4e4b\u7c7b\u7b80\u5355\u7684\u65b9\u6cd5\u6765\u5b9e\u73b0\uff0c\u800c\u9700\u8981\u7528\u5230\u795e\u7ecf\u7f51\u7edc\u3002spaCy\u9700\u8981\u4f7f\u7528\u795e\u7ecf\u7f51\u7edc\u9884\u8bad\u7ec3\u7684\u6307\u4ee3\u6d88\u89e3\u7cfb\u6570\uff0c\u5982\u679c\u524d\u9762\u6ca1\u6709\u5b89\u88c5\uff0c\u53ef\u8fd0\u884c\u547d\u4ee4\uff1apip install neuralcoref<\/li>\n<li><strong>Matcher\uff0c\u6a21\u5f0f\u5339\u914d\u3002<\/strong>\u5c31\u662f\u67e5\u627e\u6ee1\u8db3\u7279\u5b9a\u8981\u6c42\u7684\u6587\u5185\u5185\u5bb9\uff0c\u53ef\u4ee5\u662f\u5355\u8bcd\uff0c\u77ed\u8bed\uff0c\u53e5\u5b50\u4e43\u81f3\u6bb5\u843d\u3002\u4f8b\u5982\uff0c\u4f60\u60f3\u627e\u6587\u7ae0\u4e2d\u6240\u6709\u88ab\u52a8\u8bed\u53e5\u3002\u90a3\u4e48\u5c31\u53ef\u4ee5\u5199\u4e00\u4e2a\u6a21\u5f0f\uff0cpassive_pattern = [{&#8216;DEP&#8217;:&#8217;nsubjpass&#8217;},{&#8216;DEP&#8217;:&#8217;aux&#8217;,&#8217;OP&#8217;:&#8217;*&#8217;},{&#8216;DEP&#8217;:&#8217;auxpass&#8217;},{&#8216;TAG&#8217;:&#8217;VBN&#8217;}]\u5176\u4e2dDEP\u662f\u4f9d\u5b58\u5173\u7cfb\uff0c nsubjpass \u662f\u88ab\u52a8\u4e3b\u8bed\uff0c\u540e\u9762\u52a0\u4e00\u4e2a\u53ef\u80fd\u5b58\u5728\u7684\u52a9\u52a8\u8bcdaux\uff0cOP\u4ee3\u8868\u53ef\u9009\uff0c\u540e\u9762\u52a0\u88ab\u52a8\u8bcd auxpass\uff0c\u6700\u540e\u9762\u52a0\u52a8\u8bcd\u7684\u8fc7\u53bb\u5f0f\uff08\u7528Part-of-speech Tag\u4e3aVBN\u5339\u914d\uff09\u3002\u6a21\u5f0f\u5339\u914d\u5bf9\u4e8e\u5168\u6587\u67e5\u8be2\u3001 \u4fe1\u606f\u63d0\u53d6\u7b49NLP\u4efb\u52a1\u5177\u6709\u5f88\u91cd\u8981\u7684\u610f\u4e49\u3002<\/li>\n<li><strong>Word Vector\uff0c \u8bcd\u5411\u91cf\uff0c<\/strong>\u662fNLP\u6700\u91cd\u8981\u7684\u6280\u672f\uff0c\u662f\u6587\u672c\u5206\u7c7b\u3001\u4fe1\u606f\u63d0\u53d6\u7b49NLP\u4efb\u52a1\u7684\u57fa\u7840\u3002Spacy \u91cc\u9762\u4e0d\u4ec5\u53ef\u4ee5\u8bbf\u95ee\u8bcd\u5411\u91cfvector, \u8fd8\u53ef\u4ee5\u8bbf\u95ee\u53e5\u5411\u91cf span.vector \u4ee5\u53ca\u6587\u7ae0\u5411\u91cf doc.vector\u3002\u65e0\u8bba\u662fDoc, Span \u8fd8\u662f Token\uff0c\u4ed6\u4eec\u90fd\u5185\u5d4c\u76f8\u4f3c\u5ea6\u8ba1\u7b97\u51fd\u6570 similarity()\u3002\u8fd9\u4e2a\u51fd\u6570\u53ef\u4ee5\u8ba1\u7b97\u4e24\u4e2a\u5411\u91cf\u7684 cos \u8ddd\u79bb\uff0c\u4ece\u800c\u77e5\u9053\u4e24\u4e2a\u5411\u91cf\u7684\u8fd1\u4f3c\u5ea6\uff0c\u6bd4\u5982 \u3010\u732b\uff0c \u72d7\uff0c \u9999\u8549\uff0c\u82f9\u679c\u3011\uff0c\u732b\u548c\u72d7\u66f4\u76f8\u8fd1\uff0c\u9999\u8549\u548c\u82f9\u679c\u66f4\u76f8\u8fd1\u3002<\/li>\n<li><strong>Display\uff0c\u53ef\u89c6\u5316\u3002<\/strong>spaCy\u53ef\u4ee5\u628a\u5b83\u7684\u529f\u80fd\u53ef\u89c6\u5316\u51fa\u6765\u3002\u4f8b\u5982\u5bf9\u4f9d\u5b58\u5206\u6790\u7684\u53ef\u89c6\u5316\uff0c\u5bf9\u547d\u540d\u5b9e\u4f53\u8bc6\u522b\u7684\u53ef\u89c6\u5316\u7b49\u7b49\u3002<\/li>\n<\/ol>\n<p>&nbsp;<\/p>\n<p><strong>spaCy<\/strong><strong>\u76f8\u5173\u8d44\u6e90<\/strong><\/p>\n<ul>\n<li>\u514d\u8d39\u5b98\u65b9\u8bfe\u7a0b\u7684\u94fe\u63a5\uff1a\u57fa\u4e8espaCy\u7684\u9ad8\u7ea7NLP\uff08<a href=\"https:\/\/link.zhihu.com\/?target=https%3A\/\/course.spacy.io\/en\/\">https:\/\/course.spacy.io\/en\/<\/a>\uff09<\/li>\n<li>\u4e00\u7bc7\u4e0d\u9519\u7684\u535a\u5ba2\u6587\u7ae0\uff0c\u5305\u62ec\u5b89\u88c5\u8fc7\u7a0b\u548c\u5176\u4ed6Spacy\u7528\u6cd5\uff08\u5165\u95e8\u535a\u5ba2\uff09\uff1a\u4f7f\u7528Python\u4e2d\u7684spaCy\u8fdb\u884c\u81ea\u7136\u8bed\u8a00\u5904\u7406\uff08<a href=\"https:\/\/link.zhihu.com\/?target=https%3A\/\/realpython.com\/natural-language-processing-spacy-python\/\">https:\/\/realpython.com\/natural-language-processing-spacy-python\/<\/a>\uff09<\/li>\n<li>Python Spacy\u7b80\u4ecb\uff08\u89c6\u9891\uff09\u2014 \u89c6\u9891\u8bb2\u5ea7\u548c\u6559\u7a0b\uff08<a href=\"https:\/\/link.zhihu.com\/?target=https%3A\/\/realpython.com\/natural-language-processing-spacy-python\/\">https:\/\/realpython.com\/natural-language-processing-spacy-python\/<\/a>\uff09<\/li>\n<\/ul>\n<p><strong>TorchText<\/strong><\/p>\n<p>\u8fd9\u662f\u53e6\u4e00\u4e2aNLP\u9884\u5904\u7406\u5e93\uff0c\u4e3b\u8981\u7279\u70b9\u662f\u548cPytorch \u7ed3\u5408\u8f83\u597d\u3002<\/p>\n<p>TorchText\u672c\u8eab\u7684\u4e3b\u8981\u529f\u80fd\u662f\u53ef\u4ee5\u6982\u62ec\u4e3aField\uff0cDataset\u548cIterator\u8fd9\u4e09\u90e8\u5206\uff1a<\/p>\n<ul>\n<li>Dataset\uff1a\u5b9a\u4e49\u6570\u636e\u96c6\u4fe1\u606f\u3002<\/li>\n<li>Field\uff1a \u6307\u660e\u8981\u5904\u7406\u6570\u636e\u6e90\u7684\u54ea\u4e2a\u5b57\u6bb5<\/li>\n<li>Iterator\uff1a\u8fed\u4ee3\u5668\uff0c\u5904\u7406\u5b57\u6bb5\u7684\u5177\u4f53\u65b9\u6cd5<\/li>\n<\/ul>\n<p>\u901a\u8fc7\u8bbe\u7f6eField\uff0cDataset\u548cIterator\uff0cTorchText\u4e5f\u53ef\u4ee5\u5b9e\u73b0\u4ece\u5206\u8bcd\u5230\u8bcd\u5411\u91cf\u7684\u5de5\u4f5c\u3002TorchText\u8fd8\u53ef\u4ee5\u8c03\u7528spaCy\u4e2d\u7684\u4e00\u4e9b\u65b9\u6cd5\u6765\u5e2e\u52a9\u5904\u7406\u8bf8\u5982\u5206\u8bcd\u3001\u8bcd\u6027\u6807\u6ce8\u7b49\u4efb\u52a1\u3002\u5177\u4f53\u4f7f\u7528\u65f6\uff0c\u4f60\u53ef\u4ee5\u5148\u5b89\u88c5Pytorch \uff0c\u518d\u5b89\u88c5TorchText\uff0c\u6700\u540e\u5b89\u88c5spaCy. \u901a\u8fc7TorchText\u5904\u7406\u597d\u6570\u636e\uff0c\u7136\u540e\u76f4\u63a5\u5582\u7ed9Pytorch \u7684\u673a\u5668\u5b66\u4e60\u6a21\u578b\u3002<\/p>\n<ul>\n<li>\u5b98\u65b9\u7f51\u7ad9\uff1a<a href=\"https:\/\/link.zhihu.com\/?target=https%3A\/\/torchtext.readthedocs.io\/en\/latest\/\"><em>https:\/\/torchtext.readthedocs.io\/en\/latest\/<\/em><\/a><\/li>\n<li>GitHub\uff1a<a href=\"https:\/\/link.zhihu.com\/?target=https%3A\/\/github.com\/pytorch\/text\"><em>https:\/\/github.com\/pytorch\/text<\/em><\/a><\/li>\n<li>\u4f7f\u7528 Pytorch \u8fdb\u884c\u7684 BERT \u6587\u672c\u5206\u7c7b\u793a\u4f8b\uff1a<a href=\"https:\/\/link.zhihu.com\/?target=https%3A\/\/towardsdatascience.com\/bert-text-classification-using-pytorch-723dfb8b6b5b\"><em>https:\/\/towardsdatascience.com\/<\/em><\/a><\/li>\n<\/ul>\n<p>&nbsp;<\/p>\n<p>&nbsp;<\/p>\n<p>&nbsp;<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u7531\u4e8e\u9879\u76ee\u539f\u56e0\u8981\u5e94\u7528\u4e00\u4e9b\u6587\u672c\u5206\u7c7b\u7684\u6280\u672f\uff0c\u4f46<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[8],"tags":[114,146],"class_list":["post-2281","post","type-post","status-publish","format-standard","hentry","category-8","tag-114","tag-146"],"_links":{"self":[{"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/posts\/2281","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/comments?post=2281"}],"version-history":[{"count":0,"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/posts\/2281\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/media?parent=2281"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/categories?post=2281"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.divis.cn\/index.php\/wp-json\/wp\/v2\/tags?post=2281"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}