{"id":41367,"date":"2023-03-18T09:00:19","date_gmt":"2023-03-18T01:00:19","guid":{"rendered":"https:\/\/swarma.org\/?p=41367"},"modified":"2023-03-18T09:00:19","modified_gmt":"2023-03-18T01:00:19","slug":"%e8%ae%ba%e6%96%87%e6%b8%85%e5%8d%95%ef%bc%9a%e4%b8%80%e6%96%87%e6%a2%b3%e7%90%86%e5%9b%a0%e6%9e%9c%e6%8e%a8%e7%90%86%e5%9c%a8%e8%87%aa%e7%84%b6%e8%af%ad%e8%a8%80%e5%a4%84%e7%90%86%e4%b8%ad%e7%9a%84","status":"publish","type":"post","link":"https:\/\/swarma.org\/?p=41367","title":{"rendered":"\u8bba\u6587\u6e05\u5355\uff1a\u4e00\u6587\u68b3\u7406\u56e0\u679c\u63a8\u7406\u5728\u81ea\u7136\u8bed\u8a00\u5904\u7406\u4e2d\u7684\u5e94\u7528"},"content":{"rendered":"
\n
\n
\n
\n

<\/p>\n<\/section>\n

\n
\n
\n
\n
\n
\n

<\/section>\n<\/section>\n
\n
\n

\u5bfc\u8bed<\/strong><\/p>\n<\/section>\n<\/section>\n

\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\n
\n
\n

\u8fd9\u7bc7\u6587\u7ae0\u662f\u7531\u7b14\u8005\u6839\u636e\u81ea\u7136\u8bed\u8a00\u9876\u7ea7\u4f1a\u8bae\u6536\u5f55\u6709\u5173\u4e8e\u56e0\u679c\u63a8\u65ad\u7684\u6587\u732e\u6574\u7406\u800c\u6210\uff0c\u5185\u5bb9\u5305\u62ec\u8bba\u6587\u7684\u57fa\u672c\u4fe1\u606f\u4ee5\u53ca\u9605\u8bfb\u7b14\u8bb0\u3002<\/span><\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\n
\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\n
<\/strong><\/section>\n<\/section>\n
\n

\u8096\u4e4b\u4eea\uff08\u5317\u4eac\u90ae\u7535\u5927\u5b66\uff09<\/span> <\/span>| \u4f5c\u8005<\/strong><\/span><\/span><\/p>\n

\u9093\u4e00\u96ea<\/span> <\/span>| <\/strong><\/span>\u7f16\u8f91<\/strong><\/p>\n


<\/p>\n<\/section>\n


<\/section>\n

<\/section>\n

\n
\n
\n
\n
\n

<\/section>\n

<\/section>\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n
\n

1. \u4ed6\u5c71\u4e4b\u77f3<\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n
 <\/span><\/section>\n
1. Papers about Causal Inference and Language<\/span><\/section>\n
https:\/\/github.com\/causaltext\/causal-text-papers<\/span><\/section>\n
2. Causality for NLP Reading List<\/span><\/section>\n
https:\/\/github.com\/zhijing-jin\/Causality4NLP_Papers<\/span><\/section>\n
3. Causal Reading Group<\/span><\/section>\n
https:\/\/github.com\/fulifeng\/Causal_Reading_Group<\/span><\/section>\n
4. awesome-causality-algorithms<\/span><\/section>\n
https:\/\/github.com\/rguo12\/awesome-causality-algorithms<\/span><\/section>\n

<\/span><\/section>\n

<\/span><\/section>\n

\n
\n
\n
\n
\n

<\/section>\n

<\/section>\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n
\n

2. \u56e0\u679c\u5de5\u5177<\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n
 <\/span><\/section>\n
1. DoWhy: An end-to-end library for causal inference<\/span><\/section>\n
https:\/\/microsoft.github.io\/dowhy\/#dowhy-an-end-to-end-library-for-causal-inference<\/span><\/section>\n
2. Causal ML: A Python Package for Uplift Modeling and Causal Inference with ML<\/span><\/section>\n
https:\/\/github.com\/uber\/causalml<\/span><\/section>\n
3. pgmpy: Python Library for learning (Structure and Parameter), inference (Probabilistic and Causal), and simulations in Bayesian Networks.<\/span><\/section>\n
https:\/\/github.com\/pgmpy\/pgmpy<\/span><\/section>\n
4. CausalNex: A Python library that helps data scientists to infer causation rather than observing correlation.<\/span><\/section>\n
https:\/\/github.com\/quantumblacklabs\/causalnex<\/span><\/section>\n
5. CausalImpact: An R package for causal inference in time series<\/span><\/section>\n
https:\/\/github.com\/google\/CausalImpact<\/span><\/section>\n
6. CausalDiscoveryToolbox: Package for causal inference in graphs and in the pairwise settings. Tools for graph structure recovery and dependencies are included.<\/span><\/section>\n
https:\/\/github.com\/FenTechSolutions\/CausalDiscoveryToolbox<\/span><\/section>\n
7. causal-learn: Causal Discovery for Python<\/span><\/section>\n
https:\/\/github.com\/cmu-phil\/causal-learn<\/span><\/section>\n

<\/section>\n

<\/section>\n

\n
\n
\n
\n
\n

<\/section>\n

<\/section>\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n
\n

3. \u6570\u636e\u96c6<\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n
 <\/span><\/section>\n
\n
\n
\n
\n
\n
\n
\n
e-CARE: a New Dataset for Exploring Explainable Causal Reasoning<\/span><\/section>\n
https:\/\/arxiv.org\/abs\/2205.05849<\/span><\/section>\n
https:\/\/github.com\/waste-wood\/e-care<\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\u6570\u636e\u96c6\u6982\u89c8\uff1a<\/span><\/section>\n
<\/section>\n

<\/section>\n
\u6570\u636e\u96c6\u793a\u4f8b\uff1a<\/span><\/section>\n
<\/section>\n

<\/section>\n

<\/span><\/section>\n

\n
\n
\n
\n
\n

<\/section>\n

<\/section>\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n
\n

4. \u7efc\u8ff0\u7c7b\u6587\u7ae0<\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n
 <\/span><\/section>\n
\n
\n
\n
\n
\n
\n
\n
A Review of Dataset and Labeling Methods for Causality Extraction<\/span><\/section>\n
https:\/\/aclanthology.org\/2020.coling-main.133<\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\u6587\u732e\u8d21\u732e\u603b\u7ed3\u5982\u4e0b\uff1a <\/span><\/p>\n

1. \u603b\u7ed3\u4e86\u56e0\u679c\u5173\u7cfb\u7684\u5173\u7cfb\u7684\u6982\u5ff5\u5e76\u4ecb\u7ecd\u4e86\u73b0\u6709\u56e0\u679c\u5173\u7cfb\u6316\u6398\u7684\u5de5\u4f5c\uff1b <\/span><\/p>\n

2. \u9488\u5bf9\u73b0\u6709\u56e0\u679c\u5173\u7cfb\u5b9e\u9a8c\u6570\u636e\u7684\u7f3a\u9677\u548c\u4e0d\u8db3\uff0c\u5bf9\u516c\u5f00\u53ef\u7528\u7684\u6570\u636e\u96c6\u8fdb\u884c\u4e86\u603b\u7ed3\uff0c\u5e76\u4ece\u591a\u4e2a\u65b9\u9762\u8fdb\u884c\u4e86\u5206\u6790\uff1b <\/span><\/p>\n

3. \u5bf9\u73b0\u6709\u9488\u5bf9\u5e8f\u5217\u6807\u8bb0\u7684\u56e0\u679c\u7814\u7a76\u65b9\u6cd5<\/span>\uff08the causal research method of sequence labeling\uff09<\/span>\u8fdb\u884c\u4e86\u5168\u9762\u7684\u603b\u7ed3\u548c\u5206\u6790\u3002 <\/span><\/p>\n

\u503c\u5f97\u5173\u6ce8\u7684\u662f\uff0c\u4f5c\u8005\u5728\u6587\u4e2d\u5bf9\u56e0\u679c\u5173\u7cfb\u7684\u8bed\u4e49\u4f5c\u4e86\u76f8\u5e94\u7684\u5b9a\u4e49\u4e0e\u4ecb\u7ecd\uff0c\u6587\u732e\u4e2d\u5c06\u56e0\u679c\u5355\u5143<\/span>\uff08causal units\uff09<\/span>\u5b9a\u4e49\u4e3a\u4ee5\u4e0b\u56db\u7c7b\uff0c\u4f8b\u5b50\u4e2d\u7684\u3008 e1 \u3009\u8868\u793a\u539f\u56e0\uff0c\u3008 e2\u3009\u8868\u793a\u7ed3\u679c\uff1a <\/span><\/section>\n
Word\uff1a\u300c\u3008 e1 \u3009 Suicide \u3008 \/e1 \u3009 is one of the leading causes of \u3008 e2 \u3009 death \u3008 \/e2 \u3009 .\u300d <\/span><\/section>\n
Phrase\uff1a\u300c \u3008 e1 \u3009 Financial stress \u3008 \/e1 \u3009 is one of the main causes of \u3008 e2 \u3009 divorce \u3008 \/e2 \u3009.\u300d <\/span><\/section>\n
Clause\uff1a\u300c\u3008 e1 \u3009 We play with a steady beat \u3008 \/e1 \u3009 so that \u3008 e2 \u3009 dancers can follow it \u3008 \/e2 \u3009.\u300d <\/span><\/section>\n
Event\uff1a\u300c\u3008 e1 \u3009 A car traveling from Guizhou to Guangdong collided head-on with a bus \u3008 \/e1 \u3009 results the \u3008 e2 \u3009 ten people, six men and four women, including the driver, died at the scene \u3008 \/e2 \u3009 .\u300d <\/span><\/section>\n

<\/span><\/section>\n
\u540c\u65f6\u4f5c\u8005\u4e5f\u603b\u7ed3\u4e86\u5728\u82f1\u6587\u8bed\u5883\u76f8\u4e0b\u5177\u6709\u56e0\u679c\u5173\u7cfb\u7684\u8fde\u63a5\u8bcd\uff0c\u5982\u4e0b\u8868\u6240\u793a\uff1a<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n
Summary of common causal connectives in English<\/span><\/section>\n

<\/span><\/section>\n
\n
\n
\n
\n
\n
\n
\n
Causal Inference in Natural Language Processing: Estimation, Prediction, Interpretation and Beyond<\/span><\/section>\n
https:\/\/arxiv.org\/abs\/2109.00725<\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n


<\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\u8be5\u7efc\u8ff0\u6587\u732e\u5c06\u81ea\u7136\u8bed\u8a00\u5904\u7406\u4e0e\u56e0\u679c\u5173\u7cfb\u7684\u7ed3\u5408\u70b9\u5212\u5206\u6210\u4e24\u4e2a\u4e0d\u540c\u7684\u7814\u7a76\u65b9\u5411\uff0c\u5206\u522b\u662f\u56e0\u679c\u6548\u5e94\u8bc4\u4f30<\/span>\uff08estimating causal effects\uff09<\/span>\u548c\u56e0\u679c\u5173\u7cfb\u9a71\u52a8\u7684\u81ea\u7136\u8bed\u8a00\u5904\u7406\u65b9\u6cd5\u3002 <\/span><\/section>\n

\u6587\u732e\u6982\u8ff0\u5982\u4e0b\uff1a <\/span><\/p>\n

1. \u4e0d\u540c\u7c7b\u578b\u7684\u56e0\u679c\u63a8\u7406\u95ee\u9898\u53ca\u5176\u6311\u6218\uff1b <\/span><\/p>\n

2. \u9762\u5411\u6587\u672c\u6570\u636e\u548c NLP \u65b9\u6cd5\u6240\u72ec\u6709\u7684\u7edf\u8ba1\u548c\u56e0\u679c\u6311\u6218\uff1b <\/span><\/p>\n

3. \u5e94\u7528\u56e0\u679c\u5173\u7cfb\u6539\u8fdb\u81ea\u7136\u8bed\u8a00\u5904\u7406\u65b9\u6cd5\u548c\u6587\u672c\u6548\u679c\u4f30\u8ba1\u4e2d\u5b58\u5728\u7684\u95ee\u9898\u3002<\/span><\/section>\n

<\/section>\n

<\/section>\n

\n
\n
\n
\n
\n

<\/section>\n

<\/section>\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n
\n

5. \u5728NLP\u4efb\u52a1\u5efa\u6a21\u4e2d\u878d\u5408\u56e0\u679c\u5173\u7cfb<\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n
 <\/span><\/section>\n
\n
\n
\n
\n
\n
\n
\n
Uncovering Main Causalities for Long-tailed Information Extraction (EMNLP 2021)<\/span><\/section>\n
\u573a\u666f\uff1aInformation Extraction<\/span><\/section>\n
\u95ee\u9898\uff1a\u7531\u6570\u636e\u96c6\u7684\u9009\u62e9\u504f\u79fb\u5e26\u6765\u7684\u957f\u5c3e\u5206\u5e03\u53ef\u80fd\u4f1a\u5bf9\u6570\u636e\u96c6\u4ea7\u751f\u4e0d\u6b63\u786e\u7684\u5173\u8054\u5173\u7cfb<\/span><\/section>\n
https:\/\/arxiv.org\/abs\/2109.05213<\/span><\/section>\n
https:\/\/github.com\/HeyyyyyyG\/CFIE<\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\u4fe1\u606f\u62bd\u53d6\u65e8\u5728\u4ece\u975e\u7ed3\u6784\u5316\u7684\u6587\u672c\u4e2d\u63d0\u53d6\u51fa\u7ed3\u6784\u5316\u7684\u4fe1\u606f\u3002\u5728\u5b9e\u9645\u573a\u666f\u4e0b\uff0c\u6570\u636e\u96c6\u7531\u9009\u62e9\u504f\u79fb\u5e26\u6765\u7684\u957f\u5c3e\u5206\u5e03\u53ef\u80fd\u4f1a\u4f7f\u5f97\u6a21\u578b\u4e60\u5f97\u4e00\u4e9b\u53ef\u7591\u7684\u5173\u8054\u5173\u7cfb\u3002 <\/span><\/section>\n

<\/span><\/section>\n

\u672c\u6587\u732e\u63d0\u51fa\u7684 CFIE \u62df\u89e3\u51b3\u4e0a\u8ff0\u95ee\u9898\uff0c\u5177\u4f53\u8d21\u732e\u5982\u4e0b\uff1a <\/span><\/p>\n

1. \u5bf9\u591a\u79cd\u4fe1\u606f\u62bd\u53d6\u4efb\u52a1<\/span>\uff08RE\uff0cNER \u548c ED\uff09<\/span>\u6784\u5efa\u7edf\u4e00\u7684\u7ed3\u6784\u56e0\u679c\u6a21\u578b<\/span>\uff08structural causal model\uff09<\/span>\u5e76\u63cf\u8ff0\u4e0d\u540c\u53d8\u91cf\u4e4b\u95f4\u7684\u56e0\u679c\u5173\u7cfb\uff1b <\/span><\/p>\n

2. \u57fa\u4e8e\u4e0a\u8ff0\u7684\u7ed3\u6784\u56e0\u679c\u6a21\u578b\uff0c\u4f7f\u7528\u76f8\u5e94\u7684\u8bed\u8a00\u7ed3\u6784\u751f\u6210\u53cd\u4e8b\u5b9e\u6837\u672c\uff0c\u4ee5\u6b64\u5728\u63a8\u7406\u9636\u6bb5\u66f4\u597d\u7684\u8ba1\u7b97\u76f4\u63a5\u56e0\u679c\u6548\u5e94<\/span>\uff08direct causal effect\uff09<\/span>\uff1b <\/span><\/p>\n


<\/section>\n
3. \u6587\u732e\u8fdb\u4e00\u6b65\u63d0\u51fa\u4e00\u79cd\u6d88\u9664\u504f\u501a\u7684\u65b9\u6cd5\u4ee5\u63d0\u4f9b\u66f4\u5177\u9c81\u68d2\u6027\u7684\u9884\u6d4b\u3002<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n
Training and inference stages of CFIE for ED<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n
Causal effect estimation<\/span><\/section>\n

<\/span><\/section>\n

CFIE \u7684\u6a21\u578b\u8bbe\u8ba1\u5982\u4e0a\u56fe\u6240\u793a\uff0c\u5206\u6210\u5982\u4e0b\u4e94\u4e2a\u6b65\u9aa4\uff1a <\/span><\/p>\n

1. \u6784\u9020 SCM \u5e76\u8bad\u7ec3\u6a21\u578b\uff1b <\/span><\/p>\n

2. \u9884\u6d4b\u6bcf\u4e2a token \u7ed3\u679c\uff1b <\/span><\/p>\n

3. \u6cbf\u7740 1st hop \u8bed\u6cd5\u4e66\u7ed9 token \u52a0\u4e0a mask \u4ee5\u6b64\u751f\u6210\u53cd\u4e8b\u5b9e\u7684\u6837\u672c\uff1b <\/span><\/p>\n

4. \u8ba1\u7b97 Total Direct Effect\uff1b <\/span><\/p>\n

5. \u8ba1\u7b97 Main Effect\u3002<\/span><\/section>\n
\n
\n
\n
\n

<\/section>\n
\n
\n
\n
Counterfactual Generator: A Weakly-Supervised Method for Named Entity Recognition (EMNLP 2020)<\/span><\/section>\n
\u573a\u666f\uff1a\u547d\u540d\u5b9e\u4f53\u8bc6\u522b<\/span><\/section>\n
\u95ee\u9898\uff1a\u547d\u540d\u5b9e\u4f53\u8bc6\u522b\u7684\u6570\u636e\u6807\u6ce8\u662f\u4e00\u9879\u52b3\u52a8\u5bc6\u96c6\u6027\u3001\u8017\u65f6\u4e14\u6602\u8d35\u7684\u4efb\u52a1\u4e14\u300c\u5b9e\u4f53\u300d\u7684\u4e0a\u4e0b\u6587\u672c\u548c\u5b9e\u4f53\u4e4b\u95f4\u6709\u8bb8\u591a\u53ef\u7591\u7684\u5173\u8054\u5173\u7cfb<\/span><\/section>\n
https:\/\/aclanthology.org\/2020.emnlp-main.590.pdf<\/span><\/section>\n
https:\/\/github.com\/xijiz\/cfgen<\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\u8bba\u6587\u4e3b\u8981\u8d21\u732e\u5982\u4e0b\uff1a <\/span><\/p>\n

1. \u57fa\u4e8e\u56e0\u679c\u5173\u7cfb\u7684\u89c6\u89d2\u63cf\u8ff0\u4e86 NER \u6a21\u578b\u63a8\u65ad\u673a\u5236\u7684\u7406\u8bba\u57fa\u7840\uff0c\u7814\u7a76\u4e86\u6a21\u578b\u8f93\u5165\u7279\u5f81\u4e0e\u8f93\u51fa\u6807\u7b7e\u4e4b\u95f4\u5b58\u5728\u7684\u53ef\u7591\u7684\u5173\u8054\u5173\u7cfb\uff1b <\/span><\/p>\n

2. \u5728\u5bf9\u5b9e\u4f53\u8fdb\u884c\u5e72\u9884\u7684\u57fa\u7840\u4e0a\uff0c\u8bba\u6587\u5728\u6709\u9650\u7684\u89c2\u6d4b\u6837\u672c\u4e2d\u63d0\u51fa\u4e86\u4e00\u79cd\u5f31\u76d1\u7763\u7684\u547d\u540d\u5b9e\u4f53\u8bc6\u522b\u65b9\u6cd5\uff0c\u5728\u591a\u4e2a NER \u6570\u636e\u96c6\u4e0a\u8bc1\u660e\u4e86\u65b9\u6cd5\u80fd\u6709\u6548\u7684\u63d0\u5347\u6a21\u578b\u6027\u80fd\u3002<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n

Structural Causal Models (SCMs) that describes the mechanism of the NER model inference.<\/span><\/p>\n


<\/span><\/section>\n
\u4f5c\u8005\u4e3a NER \u6a21\u578b\u7684\u63a8\u7406\u673a\u5236\u6784\u5efa\u4e86\u76f8\u5e94\u7684 SCM \u5e76\u4ee5 DAG \u7684\u5f62\u5f0f\u53ef\u89c6\u5316\uff0c\u5176\u4e2d\u7ed3\u70b9 G \u8868\u793a\u6df7\u6742\u56e0\u5b50\u2014\u2014\u65e2\u5f71\u54cd\u4e86\u5b9e\u4f53\u7ed3\u70b9 E\u7684\u540c\u65f6\u4e5f\u5f71\u54cd\u4e86\u4e0a\u4e0b\u6587\u7ed3\u70b9 C\uff0c\u7ed3\u70b9 X \u5219\u4ee3\u8868\u4e86\u7531\u7ed3\u70b9 E \u548c C \u6240\u751f\u6210\u7684\u8f93\u5165\u5b9e\u4f8b\uff0c\u7ed3\u70b9 Y \u662f NER \u6a21\u578b\u7684\u8bc4\u4f30\u6307\u6807\uff0c\u5982 F1 \u503c\u3002 <\/span><\/section>\n

<\/span><\/section>\n
\u4f5c\u8005\u901a\u8fc7\u5e73\u5747\u56e0\u679c\u6548\u5e94 ACE<\/span>\uff08Average Causal Effect\uff09<\/span>\u6765\u8bc4\u4f30\u5e72\u9884\u540e\u7684 treatment effect\uff0c\u503c\u5f97\u6ce8\u610f\u7684\u662f\u4e0a\u56fe\u4e2d\u7684 (b) \u4e0e (c) \u5206\u522b\u8868\u793a\u5bf9\u4e0a\u4e0b\u6587\u548c\u5b9e\u4f53\u8fdb\u884c\u5e72\u9884\uff0c\u4f5c\u8005\u8bbe\u8ba1\u4e86\u4e00\u4e2a\u65b0\u7684\u8bc4\u4f30\u6307\u6807\u2014\u2014 RI<\/span>\uff08Relative Importance\uff09<\/span>\u6765\u8bc4\u4f30\u4e0a\u4e0b\u6587\u548c\u5b9e\u4f53\u7684\u8868\u793a\u5bf9\u4e8e NER \u6a21\u578b\u63a8\u7406\u9636\u6bb5\u7684\u91cd\u8981\u6027\uff0c\u5728\u540e\u7eed\u7684\u5b9e\u9a8c\u5206\u6790\u6a21\u5757\uff0c\u4f5c\u8005\u5f97\u51fa\u7684\u7ed3\u8bba\u662f\u5b9e\u4f53\u7684\u8868\u793a\u5bf9 NER \u6a21\u578b\u7684\u63a8\u7406\u66f4\u91cd\u8981\u3002<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n

An example of the work\ufb02ow of the Counterfactual Generator on the medical dataset.<\/span><\/p>\n


<\/span><\/section>\n

\u8bba\u6587\u4e2d\u7684\u6a21\u578b\u7684\u6d41\u7a0b\u5982\u4e0a\u56fe\u6240\u793a\uff0c\u5206\u6210\u4e09\u4e2a\u6b65\u9aa4\uff1a<\/span><\/p>\n

1. \u4ece\u5df2\u6709\u7684\u672c\u5730\u6570\u636e\u4e2d\u62bd\u51fa\u5b9e\u4f53\u96c6\u5408\uff1b<\/span><\/p>\n

2. \u4ece\u5b9e\u4f53\u96c6\u5408\u4e2d\u9009\u53d6\u4e0e\u5f85\u5e72\u9884\u6837\u672c\u5b9e\u4f53\u76f8\u540c\u7c7b\u522b\u7684\u5b9e\u4f53\u8fdb\u884c\u66ff\u6362\u4ece\u800c\u751f\u6210\u53cd\u4e8b\u5b9e\u6837\u672c\uff1b<\/span><\/p>\n

3. \u901a\u8fc7\u4f7f\u7528\u539f\u59cb\u6570\u636e\u8bad\u7ec3\u7684\u8fa8\u522b\u5668<\/span>\uff08discriminator\uff09<\/span>\u5206\u8fa8\u65b0\u751f\u6210\u7684\u53cd\u4e8b\u5b9e\u6837\u672c\u662f\u5426\u5408\u7406\uff0c\u82e5\u5408\u7406\u5219\u5c06\u8be5\u6837\u672c\u52a0\u5165\u5230\u5df2\u6709\u6570\u636e\u96c6\u5185\u3002<\/span><\/section>\n
\n
\n
\n
\n

<\/section>\n
\n
\n
\n
Counterfactual Off-Policy Training for Neural Dialogue Generation (EMNLP 2020)<\/span><\/section>\n
\u573a\u666f\uff1a\u5f00\u653e\u57df\u5bf9\u8bdd\u751f\u6210<\/span><\/section>\n
\u95ee\u9898\uff1a\u7531\u4e8e\u6f5c\u5728\u53ef\u80fd\u7684\u56de\u5e94\u6570\u91cf\u8fc7\u4e8e\u5e9e\u5927\uff0c\u5f00\u653e\u57df\u5bf9\u8bdd\u751f\u6210\u5f80\u5f80\u5904\u4e8e\u6570\u636e\u96c6\u4e0d\u8db3\u7684\u95ee\u9898\u3002<\/span><\/section>\n
https:\/\/arxiv.org\/abs\/2004.14507<\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\u8bba\u6587\u4e3b\u8981\u8d21\u732e\u5982\u4e0b\uff1a <\/span><\/p>\n

1. \u4ee5\u300c\u7ed3\u6784\u56e0\u679c\u6a21\u578b\u300d\u5bf9\u300c\u5bf9\u8bdd\u751f\u6210\u6a21\u578b\u300d\u5efa\u6a21\uff0c\u4ece\u800c\u5728\u300c\u5bf9\u8bdd\u751f\u6210\u6a21\u578b\u300d\u4e2d\u878d\u5408\u300c\u53cd\u4e8b\u5b9e\u63a8\u7406\u300d\uff1b<\/span><\/p>\n

2. \u8bba\u6587\u4e2d\u63d0\u51fa\u7684\u6a21\u578b\u6240\u751f\u6210\u7684\u53cd\u4e8b\u5b9e\u56de\u5e94\u76f8\u8f83\u4e8e\u5176\u4ed6\u6807\u51c6\u7684\u57fa\u4e8e\u5bf9\u6297\u5b66\u4e60\u4ece\u5934\u5f00\u59cb\u751f\u6210\u7684\u56de\u5e94\u7684\u8d28\u91cf\u8981\u9ad8\u5f97\u591a\uff1b<\/span><\/p>\n

3. \u8bba\u6587\u63d0\u51fa\u7684\u65b9\u6cd5\u4e0e\u6a21\u578b\u65e0\u5173<\/span>\uff08model-agnostic\uff09<\/span>\uff0c\u56e0\u6b64\u53ef\u4ee5\u9002\u914d\u4e0e\u4efb\u4f55\u57fa\u4e8e\u5bf9\u6297\u5b66\u4e60\u7684\u5bf9\u8bdd\u751f\u6210\u6a21\u578b\u3002<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n

An example of generated responses given dialogue history between person A and B.<\/span><\/p>\n


<\/span><\/section>\n

\u6587\u4e2d\u63d0\u51fa\u7684\u300ccounterfactual off-policy training (COPT) approach\u300d\u6b65\u9aa4\u5982\u4e0b\uff1a <\/span><\/p>\n

1. \u6784\u9020\u751f\u6210\u5f0f\u5bf9\u8bdd\u6a21\u578b\u7684 SCM\uff0c\u5c06\u5176\u63cf\u8ff0\u4e3a\u4e24\u4e2a\u7ec4\u6210\u90e8\u5206\u5373\u2014\u2014\u573a\u666f<\/span>\uff08scenarios\uff09<\/span>\u548c\u56e0\u679c\u673a\u5236<\/span>\uff08causal mechanisms\uff09<\/span>\uff1b <\/span><\/p>\n

2. \u7ed9\u51fa\u89c2\u6d4b\u56de\u590d\u6570\u636e\u4e2d\u63a8\u65ad\u7684\u573a\u666f\uff0cCOPT \u5c06\u573a\u666f<\/span>\uff08scenario\uff09<\/span>\u4ee5\u53ca\u5bf9\u8bdd\u5386\u53f2<\/span>\uff08dialogue history\uff09<\/span>\u6839\u636e SCM \u751f\u6210\u76f8\u5e94\u7684\u53cd\u4e8b\u5b9e\u56de\u590d\uff1b<\/span><\/p>\n

3. \u5224\u522b\u5668\u8bc4\u4f30\u751f\u6210\u7684\u8bed\u53e5\uff0c\u5e76\u5c06\u76f8\u5e94\u7684 reward \u8fd4\u8fd8\u7ed9\u7b2c\u4e8c\u6b65\u7684\u751f\u6210\u5668\u3002<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n
An example of an SCM and an intervention.<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n
The architecture of our COPT approach.<\/span><\/section>\n

<\/section>\n
\n
\n
\n
\n
\n
\n
\n
\n
Identifying Spurious Correlations for Robust Text Classi\ufb01cation (EMNLP Findings 2020)<\/span><\/section>\n
\u573a\u666f\uff1a\u6587\u672c\u5206\u7c7b<\/span><\/section>\n
\u95ee\u9898\uff1a\u6587\u672c\u5206\u7c7b\u5668\u901a\u5e38\u4f9d\u8d56\u4e8e\u53ef\u7591\u7684\u5173\u8054\u5173\u7cfb\uff0c\u901a\u5e38\u8fd9\u90e8\u5206\u5173\u7cfb\u4e0d\u4f1a\u5f71\u54cd\u6a21\u578b\u6700\u7ec8\u7684\u51c6\u786e\u6027\uff0c\u56e0\u4e3a\u5173\u8054\u5173\u7cfb\u540c\u65f6\u51fa\u73b0\u5728\u8bad\u7ec3\u96c6\u548c\u6d4b\u8bd5\u96c6\u4e2d\u3002\u4f46\u5728\u6d4b\u8bd5\u96c6\u548c\u8bad\u7ec3\u96c6\u5206\u5e03\u4e0d\u540c\uff08dataset shift\uff09 \u548c \u8bad\u7ec3\u6d4b\u8bd5\u96c6\u53ea\u5728\u5c0f\u90e8\u5206\u6837\u672c\uff08algorithmic fairness\uff09\u7684\u60c5\u51b5\u4e0b\u4f1a\u5f71\u54cd\u6a21\u578b\u51c6\u786e\u6027\u3002<\/span><\/section>\n
https:\/\/aclanthology.org\/2020.findings-emnlp.308<\/span><\/section>\n
https:\/\/github.com\/tapilab\/emnlp-2020-spurious<\/span><\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\u4f5c\u8005\u5bf9\u4e00\u4e2a\u7535\u5f71\u8bc4\u8bba\u6570\u636e\u96c6 [7] \u4f7f\u7528\u8bcd\u888b\u6a21\u578b\u903b\u8f91\u56de\u5f52\u5206\u7c7b\u5668\uff0c\u901a\u8fc7\u89c2\u6d4b\u6bcf\u4e2a\u5355\u8bcd\u7684\u7cfb\u6570\u53ef\u4ee5\u8bc4\u4f30\u8be5\u5355\u8bcd\u5bf9\u6587\u672c\u6a21\u578b\u7684\u91cd\u8981\u7a0b\u5ea6\u3002<\/span><\/section>\n
<\/section>\n

Motivating example of spurious and genuine correlations in a sentiment classi\ufb01cation task.<\/span><\/p>\n


<\/span><\/section>\n
\u5728\u4e0a\u56fe\u4e2d\uff0c\u4f5c\u8005\u5c55\u793a\u4e86\u516b\u4e2a\u4e0e\u6a21\u578b\u9ad8\u5ea6\u5339\u914d\u7684\u8bcd\u6c47\uff0c\u5176\u4e2d\u7c7b\u522b 1 \u4ee3\u8868\u6b63\u5411\u60c5\u611f\uff0c\u7c7b\u522b 2 \u5219\u8868\u793a\u8d1f\u5411\u60c5\u611f\u3002 <\/span><\/section>\n

<\/span><\/section>\n
\u4e0d\u96be\u53d1\u73b0\uff0c\u5728\u6b63\u5411\u60c5\u611f\u65b9\u9762\uff0c\u300cspielberg\u300d\u4e0e\u300canimated\u300d\u770b\u4e0a\u53bb\u8f83\u4e3a\u53ef\u7591\uff0c\u4f5c\u8005\u8ba4\u4e3a\u300cspielberg\u300d\u662f\u4e00\u540d\u975e\u5e38\u6210\u529f\u7684\u7535\u5f71\u5bfc\u6f14\uff0c\u5728\u6570\u636e\u96c6\u4e2d\u63d0\u53ca\u5230\u4ed6\u7684\u7535\u5f71\u8bc4\u8bba\u5f80\u5f80\u662f\u6b63\u5411\u7684\uff0c\u56e0\u6b64\u6a21\u578b\u8ba4\u4e3a\u300cspielberg\u300d\u4e0e\u6b63\u5411\u60c5\u611f\u5f3a\u76f8\u5173\uff0c\u4f46\u4f5c\u8005\u8ba4\u4e3a\u300cspielberg\u300d\u8fd9\u4e2a\u5355\u8bcd\u672c\u8eab\u4e0d\u5e94\u5f53\u662f\u7535\u5f71\u8bc4\u8bba\u88ab\u5224\u5b9a\u4e3a\u6b63\u5411\u60c5\u611f\u7684\u539f\u56e0\uff0c\u4e3e\u4e2a\u5f88\u7b80\u5355\u7684\u4f8b\u5b50\uff0c\u5018\u82e5\u6709\u4e00\u5929\u5bfc\u6f14\u62cd\u6444\u4e86\u4e00\u90e8\u65b0\u7247\u4f46\u53e3\u7891\u5374\u4e0d\u5c3d\u5982\u4eba\u610f\uff0c\u542b\u6709\u8fd9\u7c7b\u53ef\u7591\u7684\u5173\u8054\u5173\u7cfb\u7684\u6a21\u578b\u5219\u4f1a\u8bef\u5224\u8bc4\u8bba\u3002 <\/span><\/section>\n

<\/span><\/section>\n

\u56e0\u6b64\uff0c\u672c\u6587\u63d0\u51fa\u4e86\u4e00\u4e2a\u76d1\u7763\u6587\u672c\u5206\u7c7b\u65b9\u6cd5\u65e8\u5728\u8ba4\u51fa\u6587\u672c\u5206\u7c7b\u4e2d\u7684\u4f2a\u76f8\u5173\u6027\u548c\u771f\u76f8\u5173\u6027\uff0c\u6587\u7ae0\u63d0\u51fa\u7684\u65b9\u6cd5\u5982\u4e0b\uff1a <\/span><\/p>\n

1. \u7528\u539f\u59cb\u6570\u636e\u96c6\u8bad\u7ec3\u5206\u7c7b\u5668 f\uff1b <\/span><\/p>\n

2. \u4ece\u5206\u7c7b\u5668 f \u4e2d\u63d0\u53d6\u51fa\u4e0e\u6bcf\u4e2a\u7c7b\u522b\u5f3a\u76f8\u5173\u7684\u82e5\u5e72\u4e2a\u8bcd\u6c47\uff1b <\/span><\/p>\n

3. \u5bf9\u524d\u5e8f\u6b65\u9aa4\u63d0\u53d6\u51fa\u6765\u7684\u6bcf\u4e2a\u8bcd\u6c47\u8ba1\u7b97\u5176\u771f\u5b9e\u6216\u53ef\u7591\u7684\u53ef\u80fd\u503c\uff1b <\/span><\/p>\n

4. \u5bf9\u7b2c\u4e8c\u6b65\u4ea7\u751f\u8bcd\u6c47\u7684\u5b50\u96c6\u8fdb\u884c\u4eba\u5de5\u6807\u6ce8\uff0c\u5e76\u8bad\u7ec3\u6210\u8bcd\u6c47\u5206\u7c7b\u5668 h\uff1b <\/span><\/p>\n

5. \u5e94\u7528 h \u6807\u6ce8\u5269\u4f59\u7684\u5355\u8bcd\u53bb\u8bc4\u4f30\u4ed6\u4eec\u662f\u5426\u53ef\u7591\u3002<\/span><\/section>\n

<\/section>\n

<\/section>\n

\n
\n
\n
\n
\n

<\/section>\n

<\/section>\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n
\n

6. \u5bf9\u8bed\u8a00\u6a21\u578b\u8fdb\u884c\u56e0\u679c\u5206\u6790<\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n
 <\/span><\/section>\n
\n
\n
\n
\n
\n
\n
\n
Causal Analysis of Syntactic Agreement Mechanisms in Neural Language Models (ACL 2021)<\/span><\/section>\n
\u573a\u666f\uff1aSyntactic Agreement Mechanisms<\/span><\/section>\n
https:\/\/arxiv.org\/abs\/2106.06087<\/span><\/section>\n
https:\/\/github.com\/mattf1n\/lm-intervention<\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\u4f5c\u8005\u60f3\u5e94\u7528\u56e0\u679c\u4e2d\u4ecb\u5206\u6790<\/span>\uff08causal mediation analysis\uff09<\/span>\u6765\u89c2\u5bdf\u9884\u8bad\u7ec3\u8bed\u8a00\u6a21\u578b\u5728\u300c\u4e3b\u8c13\u4e00\u81f4\u300d\u4e0a\u7684\u8868\u73b0\uff0c\u63a2\u8ba8\u5bf9\u4e0d\u540c\u7684\u8bed\u6cd5\u7ed3\u6784\u6765\u8bf4\uff0c\u8bed\u8a00\u6a21\u578b\u4e2d\u7684\u7684\u54ea\u4e2a\u795e\u7ecf\u5143\u8d1f\u8d23\u5224\u65ad\u5224\u65ad\u300c\u4e3b\u8c13\u4e00\u81f4\u300d\u3002<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n

<\/section>\n

<\/section>\n

\n
\n
\n
\n
\n

<\/section>\n

<\/section>\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n
\n

7. \u56e0\u679c\u5173\u7cfb\u6316\u6398<\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n
 <\/span><\/section>\n
\n
\n
\n
\n
\n
\n
\n
Everything Has a Cause: Leveraging Causal Inference in Legal Text Analysis (NAACL 2021)<\/span><\/section>\n
\u573a\u666f\uff1a\u6cd5\u5f8b\u6587\u672c\u5206\u6790\uff08Legal Text Analysis\uff09<\/span><\/section>\n
\u95ee\u9898\uff1a\u73b0\u6709\u7684\u56e0\u679c\u63a8\u65ad\u7814\u7a76\u5927\u591a\u5173\u6ce8\u7ed3\u6784\u5316\u6570\u636e\uff0c\u5728\u7c7b\u4f3c\u6cd5\u5f8b\u6587\u672c\u8fd9\u6837\u7684\u975e\u7ed3\u6784\u5316\u6570\u636e\u8f83\u5c11\u88ab\u5173\u6ce8\u3002<\/span><\/section>\n
https:\/\/arxiv.org\/abs\/2104.09420<\/span><\/section>\n
https:\/\/github.com\/xxxiaol\/GCI\/<\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
<\/section>\n
An example of generated causal graph for the charge fraud.<\/span><\/section>\n

<\/span><\/section>\n
\u56e0\u679c\u63a8\u7406\u53ef\u4ee5\u6355\u6349\u4e0d\u540c\u53d8\u91cf\u4e4b\u95f4\u7684\u56e0\u679c\u5173\u7cfb\uff0c\u5927\u591a\u6570\u73b0\u6709\u5de5\u4f5c\u4e13\u6ce8\u4e8e\u89e3\u51b3\u7ed3\u6784\u5316\u6570\u636e\uff0c\u800c\u4ece\u975e\u7ed3\u6784\u5316\u6570\u636e\u4e2d\u6316\u6398\u56e0\u679c\u5173\u7cfb\u5219\u9c9c\u5c11\u6709\u5b66\u8005\u6d89\u8db3\uff0c\u672c\u6587\u732e\u63d0\u51fa\u4e86\u4e00\u4e2a\u5168\u65b0\u7684\u57fa\u4e8e\u56fe\u7684\u56e0\u679c\u63a8\u65ad\u6846\u67b6 CGI<\/span>\uff08Graph-based Causal Inference\uff09<\/span>\u2014\u2014\u4e0d\u7528\u4eba\u4e3a\u4ecb\u5165\u5373\u53ef\u901a\u8fc7\u4e8b\u5b9e\u63cf\u8ff0\u6784\u9020\u56e0\u679c\u56fe\uff0c\u4ee5\u6b64\u8f85\u52a9\u6cd5\u5f8b\u4ece\u4e1a\u8005\u4f5c\u51fa\u66f4\u597d\u7684\u5224\u65ad\u3002 <\/span><\/section>\n

<\/span><\/section>\n
\u6587\u732e\u901a\u8fc7\u300c\u8fd1\u4f3c\u6307\u63a7\u6d88\u6b67\u300d<\/span>\uff08similar charge disambiguation\uff09<\/span>\u4efb\u52a1\u6765\u8bc4\u4f30\u6846\u67b6\u7684\u6027\u80fd\uff0c\u5b9e\u9a8c\u7ed3\u679c\u9a8c\u8bc1 GCI \u4e0d\u4ec5\u80fd\u591f\u5728\u591a\u4e2a\u8fd1\u4f3c\u6307\u63a7\u7684\u4e8b\u5b9e\u63cf\u8ff0\u4e2d\u6355\u6349\u7ec6\u5fae\u5dee\u522b\uff0c\u800c\u4e14\u80fd\u63d0\u4f9b\u89e3\u91ca\u6027\u5224\u51b3\uff0c\u5728\u5c0f\u6837\u672c\u7684\u5b9e\u9a8c\u73af\u5883\u4e0b CGI \u7684\u8868\u73b0\u4f18\u5f02\u3002\u9664\u6b64\u4e4b\u5916\uff0c CGI \u4e2d\u6240\u5305\u542b\u7684\u56e0\u679c\u77e5\u8bc6\u80fd\u591f\u6709\u6548\u5730\u4e0e\u795e\u7ecf\u7f51\u7edc\u76f8\u7ed3\u5408\uff0c\u4ee5\u6b64\u63d0\u4f9b\u66f4\u597d\u7684\u6027\u80fd\u548c\u53ef\u89e3\u91ca\u6027\u3002 <\/span><\/section>\n

<\/span><\/section>\n

\u4e00\u8a00\u4ee5\u853d\u4e4b\uff0c\u6587\u732e\u7684\u4e3b\u8981\u8d21\u732e\u5982\u4e0b\u6240\u793a\uff1a <\/span><\/p>\n

1. \u6587\u732e\u63d0\u51fa\u4e86\u4e00\u4e2a\u65b0\u7684\u57fa\u4e8e\u56fe\u7684\u56e0\u679c\u63a8\u65ad\u6846\u67b6\uff0c\u53ef\u4ee5\u81ea\u52a8\u5730\u5bf9\u975e\u7ed3\u6784\u5316\u6570\u636e\u8fdb\u884c\u56e0\u679c\u63a8\u65ad\uff1b <\/span><\/p>\n

2. \u6587\u732e\u5c06 GCI \u6846\u67b6\u6240\u5f97\u51fa\u7684\u56e0\u679c\u77e5\u8bc6\u4e0e\u795e\u7ecf\u7f51\u7edc\u76f8\u7ed3\u5408\uff1b <\/span><\/p>\n

3. \u6587\u732e\u901a\u8fc7\u300c\u8fd1\u4f3c\u6307\u63a7\u6d88\u5c90\u300d\u4efb\u52a1\u9a8c\u8bc1\u4e86 CGI \u80fd\u4ece\u6cd5\u5f8b\u6587\u672c\u4e2d\u6355\u6349\u5230\u7ec6\u5fae\u5dee\u522b\uff0c\u4e14\u5b83\u80fd\u591f\u8fdb\u4e00\u6b65\u63d0\u5347\u795e\u7ecf\u7f51\u7edc\u7684\u53ef\u89e3\u91ca\u6027\u3002<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n
Overall architecture of GCI.<\/span><\/section>\n

<\/span><\/section>\n
\u672c\u6587\u732e\u63d0\u51fa\u7684 GCI \u6a21\u578b\u901a\u8fc7Yake [10] \u2014\u2014\u4e00\u79cd\u57fa\u4e8e\u6587\u672c\u7279\u5f81\u5168\u81ea\u52a8\u7684\u63d0\u53d6\u6587\u672c\u5173\u952e\u5b57\u7684\u65b9\u6cd5\u4ece\u6cd5\u5f8b\u6587\u672c\u6570\u636e\u96c6\u4e2d\u63d0\u53d6\u76f8\u5e94\u7684\u5173\u952e\u56e0\u5b50\uff0c\u8fdb\u800c\u4f7f\u7528 GFCI<\/span>\uff08Greedy Fast Causal Inference\uff09<\/span>\u7b97\u6cd5\u5bf9\u4e0a\u8ff0\u6b65\u9aa4\u63d0\u53d6\u51fa\u6765\u7684\u5173\u952e\u56e0\u5b50\u8fdb\u884c\u56e0\u679c\u53d1\u73b0\uff0c\u56e0\u4e3a GFCI \u7b97\u6cd5\u4ea7\u51fa\u7684\u7ed3\u679c\u662f\u90e8\u5206\u6709\u5411\u65e0\u73af\u56fe\u800c\u6587\u672c\uff0c\u800c\u56e0\u679c\u5173\u7cfb\u56fe\u5e94\u5f53\u662f\u6709\u5411\u65e0\u73af\u56fe\uff0c\u56e0\u6b64\u6587\u732e\u4e0b\u4e00\u90e8\u5bf9 PAG \u8fdb\u884c\u91c7\u6837\u5e76\u4f7f\u7528 ATE \u6765\u8bc4\u4f30\u56e0\u679c\u5f3a\u5ea6\uff0c\u6700\u7ec8\u5f97\u5230\u4e00\u4e2a DAG \u5206\u7c7b\u5668\u3002<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n

Two ways of integrating causal analysis and neural networks.<\/span><\/p>\n


<\/section>\n
\u6587\u732e\u901a\u8fc7\u4e24\u79cd\u65b9\u5f0f\u5c06\u56e0\u679c\u53d1\u73b0\u7684\u7ed3\u679c\u5e94\u7528\u81f3\u795e\u7ecf\u7f51\u7edc\uff0c\u7b2c\u4e00\u79cd\u662f\u5c06\u56e0\u679c\u5f3a\u5ea6\u9650\u5236 Attention \u7684\u6743\u91cd\uff0c\u7b2c\u4e8c\u79cd\u5c06\u6709\u5411\u65e0\u73af\u56fe\u62c6\u89e3\u6210\u82e5\u5e72\u56e0\u679c\u94fe\u5e76\u4f5c\u4e3a LSTM \u7684\u8f93\u5165\u3002<\/span><\/section>\n

<\/section>\n

<\/span><\/section>\n

\n
\n
\n
\n
\n

<\/section>\n

<\/section>\n
\n
<\/svg><\/section>\n<\/section>\n<\/section>\n
\n

8. \u56e0\u679c\u5e38\u8bc6\u63a8\u7406\u53ca\u5176\u751f\u6210<\/strong><\/strong><\/p>\n<\/section>\n

\n

<\/section>\n

<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/h3>\n
 <\/span><\/section>\n
\n
\n
\n
\n
\n
\n
\n
Guided Generation of Cause and Effect (IJCAI 2020)<\/span><\/section>\n
\u573a\u666f\uff1a\u6587\u672c\u751f\u6210<\/span><\/section>\n
\u95ee\u9898\uff1a\u5bf9\u4e8e\u5404\u79cd\u4eba\u5de5\u667a\u80fd\u4efb\u52a1\u6765\u8bf4\uff0c\u56e0\u679c\u77e5\u8bc6\u7684\u4e60\u5f97\u81f3\u5173\u91cd\u8981\uff0c\u4f8b\u5982\uff1a\u56e0\u679c\u56fe\u7684\u6784\u5efa\u3001\u9605\u8bfb\u7406\u89e3\u548c\u4e8b\u4ef6\u9884\u6d4b\u3002<\/span><\/section>\n
https:\/\/arxiv.org\/abs\/2107.09846<\/span><\/section>\n
https:\/\/github.com\/eecrazy\/CausalBank<\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\u8bba\u6587\u4f5c\u51fa\u5982\u4e0b\u8d21\u732e\uff1a <\/span><\/p>\n

1. \u63d0\u51fa\u4e86\u5f00\u653e\u5f0f\u56e0\u679c\u751f\u6210\u7684\u4efb\u52a1\uff1a\u5bf9\u4efb\u610f\u683c\u5f0f\u7684\u6587\u672c\u751f\u4ea7\u51fa\u5176\u53ef\u80fd\u7684\u539f\u56e0\u4e0e\u7ed3\u679c\uff1b <\/span><\/p>\n

2. \u6784\u9020\u4e86\u56e0\u679c\u6570\u636e\u96c6 CausalBank\uff0c\u5176\u5305\u542b\u6709 3.14 \u4ebf\u4e2a\u56e0\u679c\u5bf9<\/span>\uff08cause-effect pairs\uff09<\/span>\uff1b <\/span><\/p>\n

3. \u62d3\u5c55\u4e86\u8bcd\u6cd5\u9650\u5236\u7684\u89e3\u7801<\/span>\uff08lexically-constrained decoding\uff09<\/span>\uff0c\u4f7f\u5176\u652f\u6301\u6790\u53d6\u6b63\u5411\u7ea6\u675f<\/span>\uff08disjunctive positive constraints\uff09<\/span>\u3002<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n

Possible causes and effects generated by our model, conditioned on the input sentence \u201cbabies cry\u201d.<\/span><\/p>\n


<\/span><\/section>\n
<\/section>\n
Our approach for generating plausible causes and effects<\/span><\/section>\n
\n
\n
\n
\n

<\/section>\n
\n
\n
\n
GLUCOSE: GeneraLized and COntextualized Story Explanations (EMNLP 2020)<\/span><\/section>\n
\u573a\u666f\uff1a\u5e38\u8bc6\u63a8\u7406<\/span><\/section>\n
\u95ee\u9898\uff1aAI \u7cfb\u7edf\u80fd\u65e0\u6cd5\u62e5\u6709\u5e38\u8bc6\u63a8\u7406\u7684\u80fd\u529b\u662f\u56e0\u4e3a\u6709\u4e24\u4e2a\u74f6\u9888\uff1a\u4e00\u662f\u5f88\u96be\u5927\u89c4\u6a21\u7684\u83b7\u5f97\u5e38\u8bc6\u6570\u636e\u96c6\uff0c\u4e8c\u662f\u5982\u4f55\u5c06\u5e38\u8bc6\u878d\u5408\u81f3\u73b0\u6709 AI \u7cfb\u7edf\u3002<\/span><\/section>\n
https:\/\/arxiv.org\/abs\/2009.07758<\/span><\/section>\n
https:\/\/github.com\/ElementalCognition\/glucose\/<\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\u6587\u732e\u4ecb\u7ecd\u4e86 GLUCOSE \u6570\u636e\u96c6\uff0c\u5f53\u7ed9\u51fa\u4e00\u5219\u77ed\u6545\u4e8b\u548c\u6545\u4e8b\u4e2d\u7684\u4e00\u4e2a\u53e5\u5b50 X\uff0cGLUCOSE \u4ece\u5341\u4e2a\u7ef4\u5ea6\u4f1a\u6355\u6349\u4e0e X \u76f8\u5173\u7684\u56e0\u679c\u89e3\u91ca\u3002 <\/span><\/section>\n

<\/span><\/section>\n
\u8fd9\u5341\u4e2a\u7ef4\u5ea6\u542f\u53d1\u4e8e\u4eba\u7c7b\u8ba4\u77e5\u5fc3\u7406\u5b66\uff0c\u8986\u76d6\u4e86 X \u901a\u5e38\u9690\u542b\u7684\u539f\u56e0\u548c\u7ed3\u679c\uff0c\u5982\uff1a\u4e8b\u4ef6\u3001\u5730\u70b9\u3001\u6240\u6709\u7269\u7b49\u7b49\u3002<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n

Entries in the GLUCOSE dataset that explain the Gage story around the sentence X= Gage turned his bike sharply.<\/span><\/p>\n

\n
\n
\n
\n

<\/section>\n
\n
\n
\n
Counterfactual Story Reasoning and Generation (EMNLP 2019)<\/span><\/section>\n
\u573a\u666f\uff1a\u53cd\u4e8b\u5b9e\u63a8\u7406<\/span><\/section>\n
\u95ee\u9898\uff1a\u5bf9\u4e8e AI \u5b8c\u5907\u7684\u7cfb\u7edf\u6765\u8bf4\uff0c\u62e5\u6709\u53cd\u4e8b\u5b9e\u63a8\u7406\u80fd\u529b\u662f\u4e00\u4e2a\u5145\u8981\u6761\u4ef6\uff0c\u5373\u5f53\u5b9e\u9645\u53d1\u751f\u7684\u4e8b\u60c5\u6539\u53d8\u65f6\uff0c\u6a21\u578b\u80fd\u7ed9\u51fa\u76f8\u5e94\u7684\u7ed3\u679c\u3002<\/span><\/section>\n
https:\/\/arxiv.org\/abs\/1909.04076<\/span><\/section>\n
https:\/\/github.com\/qkaren\/Counterfactual-StoryRW<\/span><\/section>\n<\/section>\n<\/section>\n<\/blockquote>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
\u5728\u6587\u732e\u4e2d\uff0c\u4f5c\u8005\u63d0\u51fa\u4e86\u4e00\u4e2a\u65b0\u7684\u4efb\u52a1\uff0c\u5373\u57fa\u4e8e\u6545\u4e8b\u7406\u89e3\u548c\u751f\u6210\u7684\u53cd\u4e8b\u5b9e\u6545\u4e8b\u91cd\u5199<\/span>\uff08Counterfactual Story Rewriting\uff09<\/span>\u3002 <\/span><\/section>\n

<\/span><\/section>\n
\u4efb\u52a1\u7684\u8f93\u5165\u662f\u539f\u59cb\u6545\u4e8b\u548c\u53cd\u4e8b\u5b9e\u6761\u4ef6\uff0c\u6a21\u578b\u9700\u8981\u901a\u8fc7\u53cd\u4e8b\u5b9e\u63a8\u7406\u91cd\u5199\u6545\u4e8b\u4ee5\u4fdd\u8bc1\u903b\u8f91\u7684\u4e00\u81f4\u6027\u3002 <\/span><\/section>\n

<\/span><\/section>\n
\u4e3e\u4e2a\u4f8b\u5b50\uff0c\u5982\u4e0b\u56fe\u6240\u793a\uff0c\u5de6\u4fa7\u63d0\u4f9b\u4e86\u539f\u59cb\u6545\u4e8b\u7248\u672c\uff0cPierre \u559c\u6b22\u4e07\u5723\u8282\uff0c\u60f3\u8981\u5728\u4e07\u5723\u8282\u626e\u6f14\u5438\u8840\u9b3c\uff0c\u56e0\u6b64\u4ed6\u51c6\u5907\u4e86\u76f8\u5e94\u7684\u6253\u626e\uff0c\u4f46\u5047\u8bbe Pierre \u60f3\u8981\u626e\u6f14\u7684\u662f\u72fc\u4eba\uff0c\u901a\u8fc7\u53cd\u4e8b\u5b9e\u63a8\u7406\uff0c\u6211\u4eec\u77e5\u9053\u63a5\u4e0b\u91cc\u7684\u6545\u4e8b\u5185\u5bb9\u8981\u4e8e\u72fc\u4eba\u76f8\u7b26\u3002<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n

<\/section>\n
\u6587\u732e\u63d0\u51fa\u4e86 TIMETRAVEL \u6570\u636e\u96c6\uff0c\u62e5\u6709 29849 \u4e2a\u53cd\u4e8b\u5b9e\u4e8b\u4f8b\uff0c\u4e0b\u56fe\u662f\u6570\u636e\u96c6\u7684\u6807\u6ce8\u8fc7\u7a0b\u3002<\/span><\/section>\n

<\/span><\/section>\n
<\/section>\n
Data annotation process for the TIMETRAVEL dataset.<\/span><\/section>\n

<\/span><\/section>\n

<\/section>\n
\n
\n
\n

\u53c2\u8003\u6587\u732e<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n

\n
\n
\n
\n


<\/p>\n

\n
\n
\n
[1] A Review of dataset and labeling methods for causality extraction https:\/\/aclanthology.org\/2020.coling-main.133\/<\/span><\/section>\n
[2] Causal Inference in Natural Language Processing: Estimation, Prediction, Interpretation and Beyond https:\/\/arxiv.org\/abs\/2109.00725<\/span><\/section>\n<\/section>\n<\/section>\n<\/section>\n
[3] Uncovering Main Causalities for Long-tailed Information Extraction https:\/\/arxiv.org\/abs\/2109.05213<\/span><\/section>\n
[4] Counterfactual Generator: A Weakly-Supervised Method for Named Entity Recognition https:\/\/aclanthology.org\/2020.emnlp-main.590\/#<\/span><\/section>\n
[5] Counterfactual Off-Policy Training for Neural Dialogue Generation https:\/\/aclanthology.org\/2020.emnlp-main.276\/<\/span><\/section>\n
[6] Identifying Spurious Correlations for Robust Text Classi\ufb01cation https:\/\/arxiv.org\/pdf\/2010.02458.pdf<\/span><\/section>\n
[7] Seeing Stars: Exploiting Class Relationships for Sentiment Categorization with Respect to Rating Scales https:\/\/aclanthology.org\/P05-1015\/<\/span><\/section>\n
[8] Causal Analysis of Syntactic Agreement Mechanisms in Neural Language Models https:\/\/arxiv.org\/abs\/2106.06087<\/span><\/section>\n
[9] Everything Has a Cause: Leveraging Causal Inference in Legal Text Analysis https:\/\/arxiv.org\/abs\/2104.09420<\/span><\/section>\n
[10] Yake https:\/\/github.com\/LIAAD\/yake<\/span><\/section>\n
[11] Guided Generation of Cause and Effect https:\/\/www.ijcai.org\/Proceedings\/2020\/0502.pdf<\/span><\/section>\n
[12] GLUCOSE: GeneraLized and COntextualized Story Explanations https:\/\/arxiv.org\/pdf\/2009.07758.pdf<\/span><\/section>\n
[13] Counterfactual Story Reasoning and Generation https:\/\/arxiv.org\/pdf\/1909.04076.pdf<\/span><\/section>\n


<\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\n
\n

\uff08\u53c2\u8003\u6587\u732e\u53ef\u4e0a\u4e0b\u6ed1\u52a8<\/strong>\u67e5\u770b\uff09<\/p>\n<\/section>\n<\/section>\n


<\/section>\n

<\/section>\n

<\/section>\n
\n
\n
\n
\n
\n

\u56e0\u679c\u8868\u5f81\u5b66\u4e60\u8bfb\u4e66\u4f1a\u542f\u52a8<\/strong><\/span><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n


<\/section>\n

\u968f\u7740\u201c\u56e0\u679c\u9769\u547d\u201d\u5728\u4eba\u5de5\u667a\u80fd\u4e0e\u5927\u6570\u636e\u9886\u57df\u5f90\u5f90\u5c55\u5f00\uff0c\u4f5c\u4e3a\u8fde\u63a5\u56e0\u679c\u79d1\u5b66\u4e0e\u6df1\u5ea6\u5b66\u4e60\u6865\u6881\u7684\u56e0\u679c\u8868\u5f81\u5b66\u4e60\uff0c\u6210\u4e3a\u5907\u53d7\u5173\u6ce8\u7684\u524d\u6cbf\u65b9\u5411\u3002\u4ee5\u5f80\u7684\u6df1\u5ea6\u8868\u5f81\u5b66\u4e60\u5728\u6570\u636e\u964d\u7ef4\u4e2d\u4fdd\u7559\u4fe1\u606f\u5e76\u8fc7\u6ee4\u566a\u97f3\uff0c\u65b0\u5174\u7684\u56e0\u679c\u79d1\u5b66\u5219\u5f62\u6210\u4e86\u56e0\u679c\u63a8\u7406\u4e0e\u53d1\u73b0\u7684\u4e00\u7cfb\u5217\u65b9\u6cd5\u3002\u968f\u7740\u4e8c\u8005\u7ed3\u5408\uff0c\u56e0\u679c\u8868\u5f81\u5b66\u4e60\u6709\u671b\u50ac\u751f\u66f4\u5f3a\u5927\u7684\u65b0\u4e00\u4ee3AI\u3002<\/span><\/p>\n


<\/p>\n

\u96c6\u667a\u4ff1\u4e50\u90e8\u7ec4\u7ec7\u4ee5\u201c\u56e0\u679c\u8868\u5f81\u5b66\u4e60\u201d\u4e3a\u4e3b\u9898\u3001\u4e3a\u671f\u5341\u5468\u7684\u8bfb\u4e66\u4f1a\uff0c\u805a\u7126\u56e0\u679c\u79d1\u5b66\u76f8\u5173\u95ee\u9898\uff0c\u5171\u5b66\u5171\u7814\u76f8\u5173\u6587\u732e\u3002\u6b22\u8fce\u4ece\u4e8b\u56e0\u679c\u79d1\u5b66\u3001\u4eba\u5de5\u667a\u80fd\u4e0e\u590d\u6742\u7cfb\u7edf\u7b49\u76f8\u5173\u7814\u7a76\u9886\u57df\uff0c\u6216\u5bf9\u56e0\u679c\u8868\u5f81\u5b66\u4e60\u7684\u7406\u8bba\u4e0e\u5e94\u7528\u611f\u5174\u8da3\u7684\u5404\u754c\u670b\u53cb\u62a5\u540d\u53c2\u4e0e\u3002\u96c6\u667a\u4ff1\u4e50\u90e8\u5df2\u7ecf\u7ec4\u7ec7\u4e09\u5b63\u201c\u56e0\u679c\u79d1\u5b66\u201d\u8bfb\u4e66\u4f1a\uff0c\u5f62\u6210\u4e86\u8d85\u8fc7\u5343\u4eba\u7684\u56e0\u679c\u79d1\u5b66\u793e\u533a\uff0c\u56e0\u679c\u8868\u5f81\u5b66\u4e60\u8bfb\u4e66\u4f1a<\/a>\u662f\u5176\u7b2c\u56db\u5b63\uff0c\u73b0\u5728\u52a0\u5165\u8bfb\u4e66\u4f1a\u5373\u53ef\u53c2\u4e0e\u56e0\u679c\u793e\u533a\u5404\u7c7b\u7ebf\u4e0a\u7ebf\u4e0b\u4ea4\u6d41\u5408\u4f5c\u3002<\/span><\/p>\n


<\/span><\/p>\n

<\/span><\/span><\/a><\/p>\n


<\/span><\/section>\n
\u8be6\u60c5\u8bf7\u89c1\uff1a<\/span><\/section>\n
\u8fde\u63a5\u56e0\u679c\u79d1\u5b66\u4e0e\u6df1\u5ea6\u5b66\u4e60\u7684\u6865\u6881\uff1a\u56e0\u679c\u8868\u5f81\u5b66\u4e60\u8bfb\u4e66\u4f1a\u542f\u52a8<\/a><\/span><\/section>\n

<\/section>\n

<\/section>\n
\n
\n
\n
\n
\n

\u540eChatGPT\u8bfb\u4e66\u4f1a\u62db\u52df\u4e2d<\/strong>
<\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\n


<\/p>\n<\/section>\n

\n

2022\u5e7411\u670830\u65e5\uff0c\u4e00\u4e2a\u73b0\u8c61\u7ea7\u5e94\u7528\u7a0b\u5e8f\u8bde\u751f\u4e8e\u4e92\u8054\u7f51\uff0c\u8fd9\u5c31\u662fOpenAI\u5f00\u53d1\u7684ChatGPT\u3002\u4ece\u95ee\u7b54\u5230\u5199\u7a0b\u5e8f\uff0c\u4ece\u63d0\u53d6\u6458\u8981\u5230\u8bba\u6587\u5199\u4f5c\uff0cChatGPT\u5c55\u73b0\u51fa\u4e86\u591a\u6837\u5316\u7684\u901a\u7528\u667a\u80fd\u3002\u4e8e\u662f\uff0c\u5fae\u8f6f\u3001\u8c37\u6b4c\u3001\u767e\u5ea6\u3001\u963f\u91cc\u3001\u8baf\u98de\uff0c\u4e92\u8054\u7f51\u5927\u4f6c\u4eec\u7eb7\u7eb7\u6469\u62f3\u64e6\u638c\u51c6\u5907\u5165\u573a\u2026\u2026\u4f46\u662f\uff0c\u8bf7\u5148\u51b7\u9759\u4e00\u4e0b\u2026\u2026 \u73b0\u5728 all in \u5927\u8bed\u8a00\u6a21\u578b\u662f\u5426\u771f\u7684\u5408\u9002\uff1f\u8981\u77e5\u9053\uff0cChatGPT\u7684\u80cc\u540e\u5176\u5b9e\u5c31\u662f\u6df1\u5ea6\u5b66\u4e60+\u5927\u6570\u636e+\u5927\u6a21\u578b\uff0c\u800c\u8fd9\u4e9b\u8981\u7d20\u65e9\u57285\u5e74\u524d\u7684AlphaGo\u65f6\u671f\u5c31\u5df2\u7ecf\u5f00\u59cb\u706b\u70ed\u4e86\u30025\u5e74\u524d\u6ca1\u6709\u6293\u4f4f\u673a\u9047\uff0c\u73b0\u5728\u53c8\u51ed\u4ec0\u4e48\u53ef\u4ee5\u642d\u4e0a\u5927\u8bed\u8a00\u6a21\u578b\u8fd9\u8d9f\u5217\u8f66\u5462\uff1f<\/span><\/p>\n


<\/p>\n

\u96c6\u667a\u4ff1\u4e50\u90e8\u7279\u522b\u7ec4\u7ec7\u201c\u540e ChatGPT\u201d\u8bfb\u4e66\u4f1a<\/a>\uff0c\u7531\u5317\u5e08\u5927\u6559\u6388\u3001\u96c6\u667a\u4ff1\u4e50\u90e8\u521b\u59cb\u4eba\u5f20\u6c5f\u8001\u5e08\u8054\u5408\u8096\u8fbe\u3001\u674e\u5ae3\u7136\u3001\u5d14\u9e4f\u3001\u4faf\u6708\u6e90\u3001\u949f\u7ff0\u5ef7\u3001\u5362\u71da\u7b49\u591a\u4f4d\u8001\u5e08\u5171\u540c\u53d1\u8d77\uff0c\u65e8\u5728\u7cfb\u7edf\u6027\u5730\u68b3\u7406ChatGPT\u6280\u672f\uff0c\u5e76\u53d1\u73b0\u5176\u5f31\u70b9\u4e0e\u77ed\u677f\u3002\u672c\u7cfb\u5217\u8bfb\u4e66\u4f1a\u7ebf\u4e0a\u8fdb\u884c\uff0c2023\u5e743\u67083\u65e5\u5f00\u59cb\uff0c\u6bcf\u5468\u4e94\u665a 19:00-21:00\uff0c\u6b22\u8fce\u62a5\u540d\u4ea4\u6d41\u3002<\/span><\/p>\n


<\/span><\/p>\n

<\/span><\/a><\/p>\n<\/section>\n

\u8be6\u60c5\u8bf7\u89c1\uff1a
<\/span><\/section>\n
\u201c\u540e ChatGPT\u201d\u8bfb\u4e66\u4f1a\u542f\u52a8\uff1a\u4ece\u901a\u7528\u4eba\u5de5\u667a\u80fd\u5230\u610f\u8bc6\u673a\u5668<\/span><\/a><\/section>\n

<\/section>\n
\u62a5\u540d\u94fe\u63a5\uff1a<\/span><\/section>\n
https:\/\/pattern.swarma.org\/study_group\/23?from=wechat<\/span><\/section>\n
<\/span><\/section>\n

<\/span><\/section>\n

<\/span><\/section>\n
\n

\u63a8\u8350\u9605\u8bfb<\/span><\/strong><\/strong><\/p>\n

1. <\/strong><\/strong>\u5982\u4f55\u91cf\u5316\u8bed\u8a00\u4eb2\u5bc6\u5ea6\uff1f\u81ea\u7136\u8bed\u8a00\u5904\u7406\u5728\u8ba1\u7b97\u793e\u4f1a\u79d1\u5b66\u4e2d\u7684\u5e94\u7528<\/a><\/strong><\/span><\/section>\n
2. <\/strong>\u56e0\u679c\u63a8\u7406\u4e0e\u7a33\u5b9a\u5b66\u4e60\uff1a\u57fa\u4e8e\u79cd\u5b50\u53d8\u91cf\u7684\u7a33\u5b9a\u5b66\u4e60\u6846\u67b6<\/a><\/strong><\/span><\/strong><\/section>\n
3. <\/strong><\/strong>\u5bfb\u627e\u53ef\u843d\u5730\u7684\u56e0\u679c\u79d1\u5b66\u8303\u5f0f\uff1a\u4ece\u56e0\u679c\u63a8\u7406\u5230\u56e0\u679c\u5b66\u4e60<\/a><\/strong><\/strong><\/span><\/strong><\/strong><\/section>\n
4. \u300a\u5f20\u6c5f\u00b7\u590d\u6742\u79d1\u5b66\u524d\u6cbf27\u8bb2\u300b\u5b8c\u6574\u4e0a\u7ebf\uff01<\/strong><\/a><\/strong><\/strong><\/strong><\/section>\n
5. \u6210\u4e3a\u96c6\u667aVIP\uff0c\u89e3\u9501\u5168\u7ad9\u8bfe\u7a0b\uff0f\u8bfb\u4e66\u4f1a<\/span><\/strong><\/a><\/strong><\/strong><\/strong><\/strong><\/section>\n
6. \u52a0\u5165\u96c6\u667a\uff0c\u4e00\u8d77\u590d\u6742\uff01<\/strong><\/a><\/strong><\/strong><\/strong><\/section>\n


<\/p>\n


<\/p>\n<\/section>\n

\u70b9\u51fb\u201c\u9605\u8bfb\u539f\u6587\u201d\uff0c\u62a5\u540d\u56e0\u679c\u8868\u5f81\u8bfb\u4e66\u4f1a<\/span><\/strong><\/section>\n

<\/mp-style-type><\/p>\n<\/div>\n","protected":false},"excerpt":{"rendered":"

\u5bfc\u8bed \u8fd9\u7bc7\u6587\u7ae0\u662f\u7531\u7b14\u8005\u6839\u636e\u81ea\u7136\u8bed\u8a00\u9876\u7ea7\u4f1a\u8bae\u6536\u5f55\u6709\u5173\u4e8e\u56e0\u679c\u63a8\u65ad\u7684\u6587\u732e\u6574\u7406\u800c\u6210\uff0c\u5185\u5bb9\u5305\u62ec\u8bba\u6587\u7684\u57fa\u672c\u4fe1\u606f\u4ee5\u53ca\u9605\u8bfb\u7b14\u8bb0\u3002 \u8096\u4e4b\u4eea\uff08\u5317\u4eac\u90ae\u7535\u5927\u5b66\uff09 | \u4f5c\u8005 \u9093\u4e00\u96ea | \u7f16\u8f91 1. \u4ed6\u5c71\u4e4b\u77f3   1. Papers about Causal Inference and …<\/p>\n","protected":false},"author":0,"featured_media":41345,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[1],"tags":[],"special":[],"_links":{"self":[{"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/posts\/41367"}],"collection":[{"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=41367"}],"version-history":[{"count":0,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/posts\/41367\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=\/wp\/v2\/media\/41345"}],"wp:attachment":[{"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=41367"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=41367"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=41367"},{"taxonomy":"special","embeddable":true,"href":"https:\/\/swarma.org\/index.php?rest_route=%2Fwp%2Fv2%2Fspecial&post=41367"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}