{"id":86595,"date":"2020-05-15T13:10:53","date_gmt":"2020-05-15T05:10:53","guid":{"rendered":"http:\/\/4563.org\/?p=86595"},"modified":"2020-05-15T13:10:53","modified_gmt":"2020-05-15T05:10:53","slug":"nlp-%e5%ae%9e%e6%88%98-%e5%a6%82%e4%bd%95%e5%88%a9%e7%94%a8-nlp-%e5%88%86%e6%9e%90%e8%b4%a2%e6%8a%a5%e6%95%b0%e6%8d%ae","status":"publish","type":"post","link":"http:\/\/4563.org\/?p=86595","title":{"rendered":"NLP \u5b9e\u6218 | \u5982\u4f55\u5229\u7528 NLP \u5206\u6790\u8d22\u62a5\u6570\u636e"},"content":{"rendered":"<div>\n<div>\n<div>\n<h1>                  NLP \u5b9e\u6218 | \u5982\u4f55\u5229\u7528 NLP \u5206\u6790\u8d22\u62a5\u6570\u636e               <\/h1>\n<p> <\/p>\n<div>\n<div> <span>\u8cc7\u6df1\u5927\u4f6c : ITrecruit1 <\/span>  <span><i><\/i> 0<\/span> <\/div>\n<div> <\/div>\n<\/p><\/div>\n<\/p><\/div>\n<\/p><\/div>\n<div isfirst=\"1\"> <\/p>\n<p>\u5927\u5bb6\u597d, \u6211\u662f Lucy, FinTech \u793e\u533a\u521b\u59cb\u4eba\u3002FinTech \u793e\u533a\u662f\u4e00\u4e2a\u62e5\u6709 50,000+\u4f1a\u5458\u7684\u91d1\u878d\u79d1\u6280\u793e\u7fa4\uff0c\u65e8\u5728\u4e3a\u91d1\u878d\u79d1\u6280\u884c\u4e1a\u8d4b\u80fd\uff0c\u81f4\u529b\u4e8e\u91d1\u878d\u79d1\u6280\u884c\u4e1a\u8d44\u6e90\u5171\u4eab\u793e\u7fa4, \u6709\u91cf\u5316\u7814\u7a76\u7fa4, \u673a\u5668\u5b66\u4e60\u7fa4, \u5317\u7f8e\u7fa4, c++ \u7fa4, \u6821\u62db\u7fa4, Lucy \u66fe\u5728\u7f8e\u56fd\u9876\u5c16\u5bf9\u51b2\u57fa\u91d1\u5de5\u4f5c, \u72ec\u81ea\u7ec4\u5efa\u4e00\u652f 45 \u4eba\u4ea4\u6613\u56e2\u961f\uff0c\u6dfb\u52a0\u5fae\u4fe1: lucylj66,\u52a0\u5165\u793e\u533a, \u6512\u4eba\u8109, \u63d0\u8ba4\u77e5,\u6c42\u804c\u62db\u8058\uff01<\/p>\n<p>\u81ea\u7136\u8bed\u8a00\u5904\u7406\u662f\u8ba1\u7b97\u673a\u79d1\u5b66\u9886\u57df\u4e0e\u4eba\u5de5\u667a\u80fd\u9886\u57df\u4e2d\u7684\u4e00\u4e2a\u91cd\u8981\u65b9\u5411\uff0c\u5b83\u7814\u7a76\u80fd\u5b9e\u73b0\u4eba\u4e0e\u8ba1\u7b97\u673a\u4e4b\u95f4\u7528\u81ea\u7136\u8bed\u8a00\u8fdb\u884c\u6709\u6548\u901a\u4fe1\u7684\u5404\u79cd\u7406\u8bba\u548c\u65b9\u6cd5\u3002\u800c<strong>\u5728 NLP \u5e94\u7528\u7684\u5404\u79cd\u884c\u4e1a\u4e2d\uff0c\u91d1\u878d\u884c\u4e1a\u56e0\u5176\u4e0e\u6570\u636e\u7684\u9ad8\u5ea6\u76f8\u5173\u6027\uff0c\u6210\u4e3a NLP \u6280\u672f\u6700\u5148\u5e94\u7528\u7684\u884c\u4e1a\u4e4b\u4e00\u3002\u4eca\u5929\u8fd9\u7bc7\u6587\u7ae0\u5c06\u4e3a\u5927\u5bb6\u4ecb\u7ecd\u5982\u4f55\u5229\u7528 NLP \u5206\u6790\u8d22\u62a5\u6570\u636e\u3002<\/strong><\/p>\n<p><strong>NLP \u6280\u672f\u5728\u91d1\u878d\u6295\u8d44\u9886\u57df\u7684\u5e94\u7528\u975e\u5e38\u5e7f\u6cdb\uff0c\u6982\u62ec\u800c\u8a00\uff0c\u4e3b\u8981\u5305\u62ec\u4ee5\u4e0b\u4e09\u4e2a\u65b9\u9762\uff1a<\/strong><\/p>\n<p><strong>1. \u4fe1\u7528\u8bc4\u5206 Credit Scoring<\/strong><\/p>\n<ul>\n<li>\n<p>LenddoEFL https:\/\/www.lenddo.com\/ \u4f1a\u5e2e\u52a9\u94f6\u884c\u7b49\u91d1\u878d\u673a\u6784\u8bc4\u4ef7\u4e2a\u4eba\u4fe1\u7528\u5206\u6570\uff0c\u7279\u522b\u662f\u6ca1\u6709\u4fe1\u7528\u5386\u53f2\u7684\u7528\u6237\u3002<\/p>\n<\/li>\n<li>\n<p>\u4e3b\u8981\u7684\u65b9\u6cd5\u662f\u4e2a\u4eba\u5c06\u7f51\u4e0a\u4fe1\u606f\u62ab\u9732\u7ed9\u8fd9\u5bb6\u516c\u53f8\uff0c\u8be5\u516c\u53f8\u4f1a\u5c06\u793e\u4ea4\u7f51\u7edc\u4fe1\u606f\u3001\u6d4f\u89c8\u8bb0\u5f55\u3001\u5730\u7406\u4f4d\u7f6e\u7b49\u4fe1\u606f\u8f6c\u5316\u6210\u4fe1\u7528\u5206\u6570\uff0c\u4ece\u800c\u5bf9\u4e2a\u4eba\u4fe1\u7528\u8bc4\u7ea7\u3002<\/p>\n<\/li>\n<\/ul>\n<p><strong>2. \u6587\u672c\u60c5\u611f\u5206\u6790 Sentiment Analysis<\/strong><\/p>\n<ul>\n<li>\n<p>\u4f4d\u4e8e\u591a\u4f26\u591a\u7684\u5bf9\u51b2\u57fa\u91d1 Periscope Capital\uff0c\u4f1a\u6536\u96c6\u5927\u91cf\u7684\u63a8\u7279\u3001\u535a\u5ba2\u6570\u636e\uff0c\u5e76\u5b9e\u65f6\u5229\u7528\u795e\u7ecf\u7f51\u7edc\u7684\u6a21\u578b\u5904\u7406\uff0c\u5f97\u5230\u53ca\u65f6\u7684\u60c5\u611f\u5206\u6570\u3002\u8fd9\u4e9b\u6570\u636e\u4e4b\u540e\u4f1a\u88ab\u7efc\u5408\u6c47\u603b\uff0c\u6765\u5e2e\u52a9\u5b9e\u65f6\u6295\u8d44\u51b3\u7b56\u7684\u5236\u5b9a\u3002\u8fd9\u4e2a\u9879\u76ee\u88ab\u6536\u5f55\u5728\u4e00\u4e2a\u540d\u53eb BUZZ US Sentiment Leaders \u7684 ETF \u6307\u6570\u57fa\u91d1\u4e2d\u3002<\/p>\n<\/li>\n<li>\n<p>Sigmoidal \u662f\u4e00\u5bb6\u5229\u7528 NLP \u6280\u672f\u5728\u91d1\u878d\u9886\u57df\u63d0\u4f9b\u54a8\u8be2\u670d\u52a1\u7684\u516c\u53f8\uff0c\u901a\u8fc7\u7ed3\u5408\u5904\u7406\u5404\u79cd\u63a8\u7279\u535a\u5ba2\u6570\u636e\u5411\u6295\u8d44\u8005\u63d0\u4f9b\u5efa\u8bae\u3002<\/p>\n<\/li>\n<\/ul>\n<p><strong>3. \u667a\u80fd\u6587\u6863\u641c\u7d22 Document Search<\/strong><\/p>\n<ul>\n<li>\n<p>AlphaSense: \u901a\u8fc7\u5bf9\u91d1\u878d\u5e02\u573a\u4fe1\u606f\u5efa\u7acb\u641c\u7d22\u5f15\u64ce\uff0c\u63d0\u4f9b\u6709\u6548\u53ca\u65f6\u7684\u91d1\u878d\u4fe1\u606f\u641c\u7d22\u670d\u52a1\u3002<\/p>\n<\/li>\n<li>\n<p>Kensho\uff1a\u88ab S&amp;P Global \u6536\u8d2d\u7684\u81ea\u7136\u8bed\u8a00\u5904\u7406\u516c\u53f8\uff0c\u4e5f\u4e00\u76f4\u5728\u63d0\u4f9b\u91d1\u878d\u4fe1\u606f\u68c0\u7d22\u670d\u52a1\uff0c\u4ed6\u4eec\u5c06\u5404\u79cd\u4e3b\u9898\u7684\u65b0\u95fb\u3001\u91d1\u878d\u6587\u6863\u6c47\u603b\u5904\u7406\uff0c\u63d0\u4f9b\u7ed9\u8bf8\u5982\u9ad8\u76db\u3001JP \u6469\u6839\u7b49\u516c\u53f8\u8f85\u52a9\u5c3d\u804c\u8c03\u67e5\u3002<\/p>\n<\/li>\n<\/ul>\n<p>\u4e0b\u9762\u7ed3\u5408\u5177\u4f53\u4f8b\u5b50\uff0c\u8be6\u7ec6\u8bf4\u660e\u5982\u4f55\u5229\u7528 NLP \u6280\u672f\u5206\u6790\u4e0a\u5e02\u516c\u53f8\u8d22\u62a5\u4f1a\u8bae\uff08 earning call \uff09\u6570\u636e\u5e76\u4ece\u4e2d\u63d0\u53d6\u51b3\u7b56\u4fe1\u606f\u3002<\/p>\n<p><strong>\u4e00\u822c\u800c\u8a00\uff0c\u5bf9\u4e8e\u81ea\u7136\u8bed\u8a00\u5904\u7406\u7684\u5e94\u7528\uff0c\u4e3b\u8981\u6709\u4e09\u4e2a\u6b65\u9aa4\uff1a<\/strong><\/p>\n<ul>\n<li>\n<p><strong>\u6587\u672c\u9884\u5904\u7406 Text preprocessing<\/strong><\/p>\n<\/li>\n<li>\n<p><strong>\u6587\u672c\u7279\u5f81\u8f6c\u5316 Text to features<\/strong><\/p>\n<\/li>\n<li>\n<p><strong>\u6d4b\u8bd5\u548c\u8fed\u4ee3 Test and refinement<\/strong><\/p>\n<\/li>\n<\/ul>\n<p><strong>\u4e00\u3001\u6587\u672c\u9884\u5904\u7406<\/strong><\/p>\n<p><strong>\u6587\u672c\u9884\u5904\u7406\u4e3b\u8981\u5305\u62ec\uff1a<\/strong><\/p>\n<p><strong>\u6d88\u9664\u566a\u58f0\uff08 Noise Removal \uff09\u3001\u8bcd\u6c47\u89c4\u8303\u5316\uff08 Lexicon Normalization \uff09\u4ee5\u53ca\u5bf9\u8c61\u6807\u51c6\u5316\uff08 Object Standardization)\u3002<\/strong><\/p>\n<p><strong>1. \u566a\u58f0\u6d88\u9664<\/strong><\/p>\n<p>\u4e3b\u8981\u5c06\u6807\u70b9\u7b26\u53f7\u3001\u7f29\u5199\u5b57\u6bcd\u3001\u505c\u6b62\u8bcd\u8bed\uff08\u6bd4\u5982\u82f1\u6587\u91cc\u7684 the\uff0c\u4e2d\u6587\u91cc\u7684\u201c\u7684\u201d\uff09\u7b49\u548c\u5206\u6790\u4e3b\u9898\u65e0\u5173\u7684\u8bcd\u8bed\u79fb\u9664\u3002\u5e38\u89c1\u7684\u505a\u6cd5\u662f\u91c7\u7528\u4e00\u4e2a\u5df2\u6709\u7684\u8bcd\u6c47\u548c\u5b9e\u4f53\u8868\uff0c\u901a\u8fc7\u7b5b\u9009\u7684\u65b9\u5f0f\u5c06\u591a\u4f59\u5185\u5bb9\u79fb\u9664\u3002<\/p>\n<p>\u5728\u82f1\u6587\u4e2d\u5e38\u7528\u7684\u91d1\u878d\u8bcd\u6c47\u8868\u4e4b\u4e00\u4e3a Loughran and McDonald (2011) financial dictionary \u91d1\u878d\u8bcd\u5178\uff0c\u91cc\u9762\u5305\u62ec\u4e86\u91d1\u878d\u884c\u4e1a\u4e2d\u5e38\u89c1\u7684\u8bcd\u6c47\u4ee5\u53ca\u60c5\u611f\u8bcd\u6c47\uff0c\u5305\u62ec 80000+\u4e2a\u5173\u952e\u8bcd\uff0c350+\u4e2a\u6b63\u9762\u8bcd\u6c47\u4ee5\u53ca 2300+\u4e2a\u8d1f\u9762\u8bcd\u6c47\uff1a<\/p>\n<p>https:\/\/sraf.nd.edu\/textual-analysis\/resources\/<\/p>\n<p><strong>2. \u8bcd\u6c47\u89c4\u8303\u5316<\/strong><\/p>\n<p>\u4e3b\u8981\u662f\u8bb2\u6bcf\u4e2a\u5355\u8bcd\u7684\u4e0d\u540c\u683c\u5f0f\u89c4\u8303\u4e3a\u540c\u4e00\u79cd\u683c\u5f0f\uff0c\u4f8b\u5982\u5728\u82f1\u6587\u4e2d\u6bcf\u4e2a\u52a8\u8bcd\u90fd\u6709\u4e0d\u540c\u7684\u65f6\u6001\uff08 e.g. play, plays, played)\u3002\u4e24\u79cd\u6700\u5e38\u89c1\u7684\u65b9\u5f0f\u4e3a:<\/p>\n<ul>\n<li>\n<p><strong>Stemming \u8bcd\u5e72\u63d0\u53d6\uff1a<\/strong> \u5c06\u5355\u8bcd\u7684\u5404\u79cd\u5f62\u5f0f\u7684\u63d0\u53d6\u4e3a\u540c\u4e00\u8bcd\u5e72 \uff08\u4f8b\u5982 playing, play, played \u90fd\u8f6c\u5316\u4e3a play).<\/p>\n<\/li>\n<li>\n<p><strong>Lemmatization \u8bcd\u5f62\u8fd8\u539f\uff1a<\/strong> \u5c06\u5355\u8bcd\u539f\u6709\u7684\u5f62\u5f0f\uff0c\u4e0e\u8bcd\u5e72\u63d0\u53d6\u4e0d\u540c\u7684\u662f\uff0c\u524d\u8005\u4ea7\u751f\u7684\u8bcd\u5e72\u53ef\u80fd\u5e76\u4e0d\u662f\u5355\u8bcd\uff0c\u4f46\u662f\u8bcd\u5f62\u8fd8\u539f\u4f1a\u63d0\u53d6\u51fa\u539f\u6709\u7684\u5355\u8bcd\u3002<\/p>\n<\/li>\n<\/ul>\n<p><img decoding=\"async\" src=\"http:\/\/4563.org\/wp-content\/uploads\/2020\/05\/20200516_5ec02db69022a.jpg\" alt=\"NLP \u5b9e\u6218 | \u5982\u4f55\u5229\u7528 NLP \u5206\u6790\u8d22\u62a5\u6570\u636e\" \/><\/p>\n<p>\u5728 Python \u4e2d\u6709 nltk (Natural Language Tool Kit) package \u53ef\u4ee5\u5b9e\u73b0\u4e0a\u8ff0\u82f1\u6587\u7684\u8bcd\u6c47\u89c4\u8303\u5316\uff0c\u4f8b\u5982\u53ef\u4ee5\u901a\u8fc7\u5bfc\u5165 PortStemmer \u5b9e\u73b0\u8bcd\u5e72\u63d0\u53d6\uff1a<\/p>\n<p>from nltk.stem import PorterStemmer<br \/> porter = PorterStemmer()<br \/> print(porter.stem(&#8220;cats\u201d)) # cat<br \/> print(porter.stem(&#8220;troubling\u201d)) # trouble<\/p>\n<p>\u6216\u5229\u7528 WordNetLemmatizer \u5b9e\u73b0\u8bcd\u5f62\u8fd8\u539f\uff1a<\/p>\n<p>from nltk.stem import WordNetLemmatizer<br \/> wnl = WordNetLemmatizer()<br \/> print(wnl.lemmatize(&#8216;cars&#8217;, &#8216;n\u2019)) # \u540d\u8bcd -&gt; car<br \/> print(wnl.lemmatize(&#8216;ate&#8217;, &#8216;v&#8217;)) # \u52a8\u8bcd -&gt; eat<\/p>\n<p><strong>3. \u5bf9\u8c61\u6807\u51c6\u5316<\/strong><\/p>\n<p>\u6307\u5c06\u4e00\u4e9b\u6709\u9519\u8bef\u7684\u8bcd\u8bed\u6807\u51c6\u5316\uff0c\u6bd4\u5982 luv \u8f6c\u5316\u4e3a love \u3002\u5e38\u89c1\u7684\u65b9\u6cd5\u662f\u91c7\u7528\u4e00\u4e2a\u66f4\u5e7f\u6cdb\u7684\u8bcd\u6c47\u8868\uff08\u5305\u542b\u5f88\u591a\u5e38\u89c1\u7684\u5bb9\u6613\u6df7\u6dc6\u3001\u51fa\u9519\u7684\u8bcd\u8bed\uff09\uff0c\u7701\u4e0b\u7684\u8bcd\u8bed\u5219\u88ab\u8003\u8651\u4e3a\u566a\u58f0\u3002<\/p>\n<p><img decoding=\"async\" src=\"http:\/\/4563.org\/wp-content\/uploads\/2020\/05\/20200516_5ec02dbb57d98.jpg\" alt=\"NLP \u5b9e\u6218 | \u5982\u4f55\u5229\u7528 NLP \u5206\u6790\u8d22\u62a5\u6570\u636e\" \/><\/p>\n<p><strong>\u4e8c\u3001\u6587\u672c\u7279\u5f81\u8f6c\u6362<\/strong><\/p>\n<p>\u6587\u672c\u7279\u5f81\u8f6c\u5316\u4e3b\u8981\u5305\u62ec<strong>\u53e5\u6cd5\u89e3\u6790 (syntactical parsing) \uff0c\u7edf\u8ba1\u7279\u5f81\uff08 statistics features \uff09\u4ee5\u53ca\u8bcd\u5d4c\u5165\uff08 word embedding \uff09\u3002<\/strong><\/p>\n<p><strong>\u53e5\u6cd5\u89e3\u6790\uff08 syntactical parsing \uff09<\/strong> \u4e3b\u8981\u5206\u4e3a\u4e24\u79cd\uff0c\u8bed\u6cd5\u4f9d\u5b58(dependency grammar) \u4ee5\u53ca\u90e8\u5206\u8bed\u97f3\u6807\u8bb0\uff08 part of speech tagging, PoS)\u3002\u4ed6\u4eec\u4e24\u8005\u90fd\u662f\u7528\u6765\u5224\u65ad\u53e5\u5b50\u7684\u7ed3\u6784\u3002<\/p>\n<ul>\n<li>\n<p><strong>\u8bed\u6cd5\u4f9d\u5b58\uff1a<\/strong> \u4e3b\u8981\u662f\u5229\u7528\u4e3b\u8c13\u5bbe\u7684\u7ed3\u6784\u6765\u5224\u65ad\u5355\u8bcd\u4e4b\u95f4\u7684\u4e3b\u8981\u5173\u7cfb\uff0c\u901a\u5e38\u901a\u8fc7\u6811\u7684\u5f62\u5f0f\u6765\u5c55\u73b0\u5355\u8bcd\u4e4b\u95f4\u7684\u5173\u7cfb\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u8bed\u97f3\u6807\u8bb0 PoS\uff1a<\/strong> \u4e3b\u8981\u4f9d\u8d56\u5df2\u6709\u7684\u6807\u8bb0\u597d\u7684\u8bed\u8a00\u7ed3\u6784\uff0c\u6765\u5e2e\u52a9\u66f4\u597d\u5730\u7406\u89e3\u5355\u8bcd\u5728\u53e5\u5b50\u4e2d\u7684\u610f\u601d\u3002\uff08\u4f8b\u5982\uff0cbook\uff1abook a fligh \u4ee5\u53ca read a book \u4e2d\u7684 book \u6709\u4e0d\u540c\u7684\u610f\u601d)<\/p>\n<\/li>\n<\/ul>\n<p><strong>\u7edf\u8ba1\u7279\u5f81\uff08 statistics feature \uff09<\/strong> \u4e3b\u8981\u6307\u7684\u662f\u8bb2\u6587\u672c\u8f6c\u5316\u4e3a\u53ef\u4ee5\u91cf\u5316\u7684\u5355\u8bcd\u4e4b\u540e\u7684\u7edf\u8ba1\u7ed3\u679c\u3002<strong>\u5e38\u89c1\u7684\u7edf\u8ba1\u7279\u5f81\u5305\u62ec\u8bcd\u6c47\u6570\u91cf\u3001\u53e5\u5b50\u6570\u91cf\u4ee5\u53ca\u97f3\u8282\u7edf\u8ba1\u3002\u901a\u8fc7\u8fd9\u4e9b\u6570\u91cf\uff0c\u53ef\u4ee5\u5b9a\u91cf\u4e00\u53e5\u8bdd\u7684\u60c5\u611f\u8272\u5f69\u3002<\/strong><\/p>\n<p><strong>\u5728\u8fd9\u4e2a\u5b9e\u4f8b\u4e2d\uff0c\u6211\u4eec\u901a\u8fc7\u8ba1\u7b97\u901a\u8fc7\u8d1f\u9762\u8bcd\u8bed\u6570\u91cf\u5360\u603b\u53e5\u5b50\u5355\u8bcd\u6570\u91cf\u7684\u6bd4\u4f8b\u6765\u4f53\u73b0\u8fd9\u53e5\u8bdd\u7684\u60c5\u611f\u8272\u5f69\u3002<\/strong><\/p>\n<p><strong>\u8bcd\u5d4c\u5165\uff08 word embedding \uff09<\/strong> \u662f\u4e00\u79cd\u66f4\u52a0\u9ad8\u7ea7\u7684\u7edf\u8ba1\u7279\u5f81\uff0c\u4e3b\u8981\u8bb2\u6587\u672c\u4e2d\u7684\u5355\u8bcd\u901a\u8fc7\u5411\u91cf\u6765\u8868\u793a\u3002\u4f8b\u5982\u4e0b\u8868\u4e2d\u6211\u4eec\u53ef\u4ee5\u627e\u51fa\u6bcf\u8bcd\u6c47\u4e4b\u95f4\u5728\u6587\u672c\u4e2d\u51fa\u73b0\u7684\u5173\u7cfb\uff0c\u7528\u4e00\u4e2a\u5728 0-1 \u4e4b\u95f4\u7684\u6570\u5b57\u8868\u793a\u4e8c\u8005\u7684\u5173\u7cfb\uff0c1 \u4ee3\u8868\u4e24\u8005\u5173\u7cfb\u975e\u5e38\u5bc6\u5207\uff0c0 \u4ee3\u8868\u6ca1\u6709\u5173\u8054\u3002<\/p>\n<p>\u5982\u4e0b\u8868\u4e2d\u56fd\u738b\uff08 king \uff09\u548c\u738b\u6743\uff08 Royalty \uff09\u7684\u5173\u7cfb\u662f 0.99 \uff0c \u800c\u5973\u738b\uff08 queen \uff09\u548c\u7537\u6027\u4e3b\u4e49\uff08 Masculinity \uff09\u7684\u5173\u7cfb\u4e0d\u5927\uff0c\u53ea\u6709 0.05 \u3002<\/p>\n<p><img decoding=\"async\" src=\"http:\/\/4563.org\/wp-content\/uploads\/2020\/05\/20200516_5ec02dc994bfe.jpg\" alt=\"NLP \u5b9e\u6218 | \u5982\u4f55\u5229\u7528 NLP \u5206\u6790\u8d22\u62a5\u6570\u636e\" \/><\/p>\n<p>\u901a\u8fc7\u6269\u5c55\u7ef4\u5ea6\uff0c\u6211\u4eec\u80fd\u591f\u5c06\u5355\u8bcd\u7528\u591a\u4e2a\u7ef4\u5ea6\u7684\u5411\u91cf\u8868\u793a\uff0c\u6700\u7ec8\u901a\u8fc7\u77e2\u91cf\u8fd0\u7b97\u627e\u51fa\u4e0d\u540c\u5355\u8bcd\u4e4b\u95f4\u7684\u542b\u4e49\uff0c\u4f8b\u5982\u5728\u4e0a\u8ff0\u4f8b\u5b50\u4e2d\uff1a\u56fd\u738b\uff08 King \uff09-\u7537\u6027\u4e3b\u4e49\uff08 masculinity \uff09+\u5973\u6027\u4e3b\u4e49\uff08 femininity \uff09=\u5973\u738b\uff08 Queen \uff09\u3002<\/p>\n<p><strong>\u4e09\u3001\u6d4b\u8bd5\u548c\u8fed\u4ee3<\/strong><\/p>\n<p>\u901a\u8fc7\u5206\u79bb\u8bad\u7ec3\u6570\u636e\u548c\u6d4b\u8bd5\u6570\u636e\uff0c\u6211\u4eec\u53ef\u4ee5\u901a\u8fc7\u8bad\u7ec3\u6570\u636e\u5e26\u5165\u5230 NLP \u7b97\u6cd5\u5f97\u51fa\u5bf9\u5e94\u7684\u6a21\u578b\uff0c\u5e76\u901a\u8fc7\u6d4b\u8bd5\u6570\u636e\u5bf9\u5176\u8fdb\u884c\u6821\u6b63\uff08 calibration \uff09\u548c\u6d4b\u8bd5\uff08 test \uff09\uff0c\u6700\u7ec8\u5f97\u5230\u7406\u60f3\u7684\u6a21\u578b\u548c\u7ed3\u679c\u3002<\/p>\n<p>\u8fd9\u91cc\u901a\u8fc7\u4e24\u4e2a\u4f8b\u5b50\u6765\u5c55\u793a NLP \u6280\u672f\u5728\u5206\u6790\u8d22\u62a5\u6570\u636e\u4e2d\u7684\u5e94\u7528\u3002<\/p>\n<p><strong>1. \u5bf9\u5355\u53ea\u80a1\u7968\u7684\u5206\u6790<\/strong><\/p>\n<p>\u5206\u6790\u82f9\u679c\u6bcf\u4e2a\u5b63\u5ea6\u7684\u8d22\u62a5\u4f1a\u8bae\u6587\u5b57\u7684\u540c\u6bd4\u60c5\u611f\u53d8\u5316\u767e\u5206\u6bd4\u548c\u4e0b\u4e00\u4e2a\u5b63\u5ea6\u80a1\u4ef7\u56de\u62a5\u4e4b\u95f4\u7684\u5173\u7cfb\u3002<\/p>\n<p>\u5728\u8fd9\u4e2a\u4f8b\u5b50\u4e2d\uff0c\u60c5\u611f\u53d8\u5316\u767e\u5206\u6bd4\u4e3a\u8fd9\u4e2a\u5b63\u5ea6\u7684\u60c5\u611f\u503c\u4e0e\u53bb\u5e74\u76f8\u540c\u5b63\u5ea6\u60c5\u611f\u503c\u53d8\u5316\u7684\u767e\u5206\u6bd4\uff08 Quarter-over-quarter, QoQ \uff09\uff0c\u60c5\u611f\u503c\u5b9a\u4e49\u4e3a\u5728\u8d22\u62a5\u4f1a\u8bae\u6587\u672c\u4e2d\u8d1f\u9762\u8bcd\u6c47\u5360\u6574\u4e2a\u6587\u672c\u7684\u6bd4\u4f8b\u3002<\/p>\n<p><img decoding=\"async\" src=\"http:\/\/4563.org\/wp-content\/uploads\/2020\/05\/20200516_5ec02dcef0edf.jpg\" alt=\"NLP \u5b9e\u6218 | \u5982\u4f55\u5229\u7528 NLP \u5206\u6790\u8d22\u62a5\u6570\u636e\" \/><\/p>\n<p>\u4ece\u4e0a\u56fe\u4e2d\u6211\u4eec\u53ef\u4ee5\u770b\u5230\uff0c\u5728 2013 \u5e74 1 \u6708\u7684\u8d22\u62a5\u4f1a\u8bae\u4e2d\u82f9\u679c\u7684\u8d1f\u9762\u60c5\u611f\u503c\u540c\u6bd4\u589e\u52a0 67%\uff0c\u800c\u5bf9\u5e94\u4e0b\u4e2a\u5b63\u5ea6\u5373 2013 \u5e74 4 \u6708\u82f9\u679c\u7684\u56de\u62a5\u7387\u4e3a-21%\u3002<\/p>\n<p>\u6574\u4f53\u800c\u8a00\uff0c\u4e24\u8005\u7684\u76ae\u5c14\u900a\u76f8\u5173\u6307\u6570\u4e3a-0.3\uff0c\u53ef\u4ee5\u770b\u51fa\u5386\u53f2\u4e0a\u82f9\u679c\u8d22\u62a5\u4f1a\u8bae\u8d1f\u9762\u60c5\u7eea\u53d8\u5316\u548c\u4e0b\u4e2a\u5b63\u5ea6\u7684\u6c47\u62a5\u6210\u8d1f\u76f8\u5173\u3002<\/p>\n<p><strong>2. \u5bf9\u6574\u4e2a\u884c\u4e1a\u7684\u5206\u6790<\/strong><\/p>\n<p>\u7b2c\u4e8c\u4e2a\u4f8b\u5b50\u53ef\u4ee5\u7528\u8fc7\u70ed\u5ea6\u56fe\u7684\u65b9\u5f0f\u6765\u5c55\u73b0\u4e0d\u540c\u7684\u884c\u4e1a\u5728\u60c5\u611f\u503c\u7684\u53d8\u5316\u3002<\/p>\n<p><img decoding=\"async\" src=\"http:\/\/4563.org\/wp-content\/uploads\/2020\/05\/20200516_5ec02dd32cd67.jpg\" alt=\"NLP \u5b9e\u6218 | \u5982\u4f55\u5229\u7528 NLP \u5206\u6790\u8d22\u62a5\u6570\u636e\" \/><\/p>\n<p>\u901a\u8fc7\u5bf9\u6bd4\u884c\u4e1a\u60c5\u611f\u503c\u53d8\u5316\uff0c\u6211\u4eec\u53ef\u4ee5\u770b\u51fa\u54ea\u4e9b\u884c\u4e1a\u60c5\u611f\u503c\u53d8\u5316\u4e3a\u6b63\u5411\uff0c\u54ea\u4e9b\u60c5\u611f\u503c\u53d8\u5316\u4e3a\u8d1f\u5411\u3002\u60c5\u611f\u503c\u63d0\u5347\u7684\u884c\u4e1a\u4ee3\u8868\u4e86\u516c\u53f8\u9ad8\u7ba1\u5bf9\u4e8e\u8be5\u884c\u4e1a\u8fd9\u4e2a\u5b63\u5ea6\u6709\u4fe1\u5fc3\uff0c\u60c5\u611f\u503c\u4e3a\u51cf\u5c11\u7684\u884c\u4e1a\u5219\u76f8\u53cd\u3002<\/p>\n<p>\u5bf9\u6bd4\u5206\u6790\u5355\u72ec\u884c\u4e1a\uff0c\u6211\u4eec\u4e5f\u53ef\u4ee5\u770b\u51fa\u54ea\u4e9b\u884c\u4e1a\u5728\u54ea\u4e9b\u65f6\u95f4\u51fa\u73b0\u4e86\u62d0\u70b9\u3002<\/p>\n<p>\u4f8b\u5982\u94f6\u884c\u4e1a Banks \u4ece 2016 \u5e74\u7b2c\u56db\u5b63\u5ea6\u4e4b\u540e\uff0c\u76f8\u6bd4\u4e4b\u524d\u5b63\u5ea6\u5747\u4e3a\u6b63\u5411\u60c5\u611f\u53d8\u5316\uff0c\u8868\u660e\u9ad8\u7ba1\u5bf9\u8be5\u884c\u4e1a\u672a\u6765\u666e\u904d\u66f4\u6709\u4fe1\u5fc3\u3002<\/p>\n<p>\u4e0b\u9762\u662f\u5229\u7528 nltk package \u5efa\u7acb\u6a21\u578b\u7684\u5173\u952e\u4ee3\u7801:<\/p>\n<p><img decoding=\"async\" src=\"http:\/\/4563.org\/wp-content\/uploads\/2020\/05\/20200516_5ec02dd6a9eee.jpg\" alt=\"NLP \u5b9e\u6218 | \u5982\u4f55\u5229\u7528 NLP \u5206\u6790\u8d22\u62a5\u6570\u636e\" \/><\/p>\n<p><img decoding=\"async\" src=\"http:\/\/4563.org\/wp-content\/uploads\/2020\/05\/20200516_5ec02de338da2.jpg\" alt=\"NLP \u5b9e\u6218 | \u5982\u4f55\u5229\u7528 NLP \u5206\u6790\u8d22\u62a5\u6570\u636e\" \/><\/p>\n<p>\u6765\u6e90: S&amp;P Global Quantamental Research<\/p>\n<p><strong>#\u4eca\u65e5\u4e92\u52a8#<\/strong><\/p>\n<p><strong>\u5173\u4e8e NLP \u7684\u5e94\u7528<\/strong><\/p>\n<p><strong>\u4f60\u6709\u6ca1\u6709\u5b9e\u6218\u7ecf\u9a8c\u6216\u60f3\u8981\u5206\u4eab\u7684\uff1f\u6b22\u8fce\u7559\u8a00\u8ba8\u8bba~<\/strong><\/p>\n<p>\u8fd1\u671f\u70ed\u62db:\uff08\u70b9\u51fb\u6807\u9898\uff0c\u5373\u53ef\u4e86\u89e3\u8be6\u60c5\uff09<\/p>\n<p><strong>1.<\/strong><strong>\u91d1\u878d\u79d1\u6280\u62db\u8058 | \u6280\u672f\u4e13\u573a<\/strong><\/p>\n<p><strong>2.<\/strong><strong>\u91d1\u878d\u79d1\u6280\u62db\u8058 | \u91cf\u5316\u4e13\u573a<\/strong><\/p>\n<p><strong>3.<\/strong><strong>\u91d1\u878d\u79d1\u6280\u62db\u8058 | \u673a\u5668\u5b66\u4e60\u4e13\u573a<\/strong><\/p>\n<p><strong>4.<\/strong><strong>\u62db\u8058 | C++ Developer &#8211; 80-100 \u4e07-\u4e0a\u6d77-\u591a\u5bb6\u5bf9\u51b2\u57fa\u91d1<\/strong><\/p>\n<p><strong>5.<\/strong><strong>\u62db\u8058 | \u6570\u636e\u5f00\u53d1\u5de5\u7a0b\u5e08-40-48 \u4e07+\u5956\u91d1-\u5317\u4eac-\u5bf9\u51b2\u57fa\u91d1<\/strong><\/p>\n<\/p><\/div>\n<div> <b>\u5927\u4f6c\u6709\u8a71\u8aaa<\/b> (<span>0<\/span>)        <\/div>\n<div> <\/div>\n<\/p><\/div>\n<\/p><\/div>\n<ul>\n<li>\n","protected":false},"excerpt":{"rendered":"<p>NLP \u5b9e\u6218 | \u5982\u4f55\u5229\u7528 NLP&hellip;<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[],"tags":[],"_links":{"self":[{"href":"http:\/\/4563.org\/index.php?rest_route=\/wp\/v2\/posts\/86595"}],"collection":[{"href":"http:\/\/4563.org\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/4563.org\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/4563.org\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/4563.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=86595"}],"version-history":[{"count":0,"href":"http:\/\/4563.org\/index.php?rest_route=\/wp\/v2\/posts\/86595\/revisions"}],"wp:attachment":[{"href":"http:\/\/4563.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=86595"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/4563.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=86595"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/4563.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=86595"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}