| { |
| "paper_id": "O18-1023", |
| "header": { |
| "generated_with": "S2ORC 1.0.0", |
| "date_generated": "2023-01-19T08:09:48.374486Z" |
| }, |
| "title": "Combining Convolutional Neural Network and Recurrent Neural Network for Tweet Polarity Classification", |
| "authors": [ |
| { |
| "first": "Chih-Ting", |
| "middle": [], |
| "last": "\u8449\u81f4\u5ef7", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Information Engineering National Sun Yat-sen University", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Yeh", |
| "suffix": "", |
| "affiliation": { |
| "laboratory": "", |
| "institution": "Information Engineering National Sun Yat-sen University", |
| "location": {} |
| }, |
| "email": "" |
| }, |
| { |
| "first": "Chia-Ping", |
| "middle": [], |
| "last": "\u9673\u5609\u5e73", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "" |
| }, |
| { |
| "first": "", |
| "middle": [], |
| "last": "Chen", |
| "suffix": "", |
| "affiliation": {}, |
| "email": "cpchen@mail.cse.nsysu.edu.tw" |
| } |
| ], |
| "year": "", |
| "venue": null, |
| "identifiers": {}, |
| "abstract": "With the development of the Internet, male and female, old and young, often use social network to share the trivia of everyday things and comment on current affairs. The amount of information generated every day is very considerable. If we analyze those data to get the impressions from society, we can easier to make better decisions. This paper chooses Twitter as the research subject and conduct sentiment analysis on English tweets. We use Tweepy to", |
| "pdf_parse": { |
| "paper_id": "O18-1023", |
| "_pdf_hash": "", |
| "abstract": [ |
| { |
| "text": "With the development of the Internet, male and female, old and young, often use social network to share the trivia of everyday things and comment on current affairs. The amount of information generated every day is very considerable. If we analyze those data to get the impressions from society, we can easier to make better decisions. This paper chooses Twitter as the research subject and conduct sentiment analysis on English tweets. We use Tweepy to", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Abstract", |
| "sec_num": null |
| } |
| ], |
| "body_text": [ |
| { |
| "text": "collect tweets on Twitter and use them to train word vector. After that, the trained word vector is fine-tuned to have emotional features by Convolutional Neural Network (CNN) .", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "Then, the fine-tuned vector is used for training in the Recurrent Neural Network (RNN) to get the final polarity classification results. Our system uses the dataset of the subtask V-oc of SemEval-2018 Task1: Affect in Tweets for training. Compared to the results of competition, we are in the fifth place. ", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| }, |
| { |
| "text": "( ) T softmax M \uf061 \uf077 = (2) T r X \uf061 = (3) \u5047\u8a2d\u8f38\u5165\u9577\u5ea6\u70ba N \uff0cd \u70ba\u7b2c\u4e8c\u5c64 Bi-LSTM \u8f38\u51fa\u7684\u7dad\u5ea6\u3001 dN X \uf0b4 \uf0ce \u70ba\u7b2c\u4e8c\u5c64 Bi-LSTM \u5404 \u6642\u9593\u7684\u8f38\u51fa [ 1 ,..., N xx ] \u6240\u7d44\u6210\u7684\u77e9\u9663\u3001 dN M \uf0b4 \uf0ce \u3001 dd att W \uf0b4 \uf0ce \u3001 d \uf077 \uf0ce \u3001 N \uf061 \uf0ce \u3001 d r \uf0ce \u3002 X \u5148\u7d93\u904e\u4e00\u5c64\u7684\u5168\u9023\u63a5\u5c64\u8f49\u63db\uff0c\u4e14\u70ba\u4e86\u4fdd\u7559\u8ca0\u503c\u9078\u64c7 tanh \u70ba\u6fc0\u6d3b\u51fd\u6578\uff0c\u63a5\u8457\u518d\u4e58\u4ee5 T \uf077 \u5f8c\u505a softmax\uff0c\u5f97\u51fa\u53e5\u5b50\u4e2d\u5404\u4f4d\u7f6e\u7684\u6b0a\u91cd\uff0c\u53e5\u5b50\u4e2d\u8f03\u91cd\u8981\u80fd\u5206\u8fa8\u60c5\u7dd2\u7684\u5730\u65b9\u6703\u6709\u8f03\u9ad8\u7684 \u6b0a\u91cd\uff0c\u6700\u5f8c\u5c07 X \u4e58\u4ee5\u6ce8\u610f\u529b\u6b0a\u91cd\uff0c\u5373\u53ef\u5f97\u5230\u53e5\u5b50\u7d93\u6ce8\u610f\u529b\u6a5f\u5236\u6574\u5408\u5f8c\u7684\u8f38\u51fa", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| } |
| ], |
| "back_matter": [], |
| "bib_entries": { |
| "BIBREF0": { |
| "ref_id": "b0", |
| "title": "Efficient estimation of word representations in vector space", |
| "authors": [ |
| { |
| "first": "T", |
| "middle": [], |
| "last": "Mikolov", |
| "suffix": "" |
| }, |
| { |
| "first": "K", |
| "middle": [], |
| "last": "Chen", |
| "suffix": "" |
| }, |
| { |
| "first": "G", |
| "middle": [], |
| "last": "Corrado", |
| "suffix": "" |
| }, |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Dean", |
| "suffix": "" |
| } |
| ], |
| "year": 2013, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "T. Mikolov, K. Chen, G. Corrado, and J. Dean, \"Efficient estimation of word representations in vector space,\" arXiv preprint arXiv1301.3781, 2013.", |
| "links": null |
| }, |
| "BIBREF1": { |
| "ref_id": "b1", |
| "title": "Glove: Global vectors for word representation", |
| "authors": [ |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Pennington", |
| "suffix": "" |
| }, |
| { |
| "first": "R", |
| "middle": [], |
| "last": "Socher", |
| "suffix": "" |
| }, |
| { |
| "first": "C", |
| "middle": [], |
| "last": "Manning", |
| "suffix": "" |
| } |
| ], |
| "year": 2014, |
| "venue": "Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP)", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "J. Pennington, R. Socher, and C. Manning. \"Glove: Global vectors for word representation,\" Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP), 2014.", |
| "links": null |
| }, |
| "BIBREF2": { |
| "ref_id": "b2", |
| "title": "Bag of tricks for efficient text classification", |
| "authors": [ |
| { |
| "first": "A", |
| "middle": [], |
| "last": "Joulin", |
| "suffix": "" |
| }, |
| { |
| "first": "E", |
| "middle": [], |
| "last": "Grave", |
| "suffix": "" |
| }, |
| { |
| "first": "P", |
| "middle": [], |
| "last": "Bojanowski", |
| "suffix": "" |
| }, |
| { |
| "first": "T", |
| "middle": [], |
| "last": "Mikolov", |
| "suffix": "" |
| } |
| ], |
| "year": 2016, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": { |
| "arXiv": [ |
| "arXiv:1607.01759" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "A. Joulin, E. Grave, P. Bojanowski, and T. Mikolov, \"Bag of tricks for efficient text classification,\" arXiv preprint arXiv:1607.01759, 2016", |
| "links": null |
| }, |
| "BIBREF3": { |
| "ref_id": "b3", |
| "title": "Senti17 at SemEval-2017 Task 4: Ten Convolutional Neural Network Z39.50 is protocol for", |
| "authors": [ |
| { |
| "first": "H", |
| "middle": [], |
| "last": "Hamden", |
| "suffix": "" |
| } |
| ], |
| "year": null, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "H. Hamden, \"Senti17 at SemEval-2017 Task 4: Ten Convolutional Neural Network Z39.50 is protocol for ...", |
| "links": null |
| }, |
| "BIBREF4": { |
| "ref_id": "b4", |
| "title": "Figure 1. The Dynamite Voters for Tweet Polarity Classification", |
| "authors": [], |
| "year": 2017, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": { |
| "arXiv": [ |
| "arXiv:1705.02023" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "Figure 1. The Dynamite Voters for Tweet Polarity Classification,\" arXiv preprint arXiv:1705.02023, 2017.", |
| "links": null |
| }, |
| "BIBREF5": { |
| "ref_id": "b5", |
| "title": "Datastories at semeval-2017 task 4: Deep lstm with attention for message-level and topic-based sentiment analysis", |
| "authors": [ |
| { |
| "first": "C", |
| "middle": [], |
| "last": "Baziotis", |
| "suffix": "" |
| }, |
| { |
| "first": "N", |
| "middle": [], |
| "last": "Pelekis", |
| "suffix": "" |
| }, |
| { |
| "first": "C", |
| "middle": [], |
| "last": "Doulkeridis", |
| "suffix": "" |
| } |
| ], |
| "year": 2017, |
| "venue": "Proceedings of the 11th International Workshop on Semantic Evaluation", |
| "volume": "", |
| "issue": "", |
| "pages": "747--754", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "C. Baziotis, N. Pelekis, and C. Doulkeridis. \"Datastories at semeval-2017 task 4: Deep lstm with attention for message-level and topic-based sentiment analysis,\" In: Proceedings of the 11th International Workshop on Semantic Evaluation (SemEval-2017), pp. 747-754, 2017.", |
| "links": null |
| }, |
| "BIBREF6": { |
| "ref_id": "b6", |
| "title": "Learning to forget: Continual prediction with LSTM", |
| "authors": [ |
| { |
| "first": "F", |
| "middle": [ |
| "A" |
| ], |
| "last": "Gers", |
| "suffix": "" |
| }, |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Schmidhuber", |
| "suffix": "" |
| }, |
| { |
| "first": "F", |
| "middle": [], |
| "last": "Cummins", |
| "suffix": "" |
| } |
| ], |
| "year": 1999, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "F.A. Gers, J. Schmidhuber, and F. Cummins,\" Learning to forget: Continual prediction with LSTM\", 1999.", |
| "links": null |
| }, |
| "BIBREF7": { |
| "ref_id": "b7", |
| "title": "BB_twtr at SemEval-2017 Task 4: Twitter Sentiment Analysis with CNNs and LSTMs", |
| "authors": [ |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Clich\u00e9", |
| "suffix": "" |
| } |
| ], |
| "year": 2017, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": { |
| "arXiv": [ |
| "arXiv:1704.06125" |
| ] |
| }, |
| "num": null, |
| "urls": [], |
| "raw_text": "M. Clich\u00e9, \"BB_twtr at SemEval-2017 Task 4: Twitter Sentiment Analysis with CNNs and LSTMs,\" arXiv preprint arXiv:1704.06125, 2017.", |
| "links": null |
| }, |
| "BIBREF8": { |
| "ref_id": "b8", |
| "title": "Proceedings of The 12th International Workshop on Semantic Evaluation", |
| "authors": [ |
| { |
| "first": "S", |
| "middle": [], |
| "last": "Mohammad", |
| "suffix": "" |
| }, |
| { |
| "first": "F", |
| "middle": [], |
| "last": "Bravo-Marquez", |
| "suffix": "" |
| }, |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Salameh", |
| "suffix": "" |
| } |
| ], |
| "year": 2018, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "1--17", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "S. Mohammad, F. Bravo-Marquez, and M. Salameh, \"Semeval-2018 task 1: Affect in tweets,\" In: Proceedings of The 12th International Workshop on Semantic Evaluation, pp. 1-17, 2018.", |
| "links": null |
| }, |
| "BIBREF9": { |
| "ref_id": "b9", |
| "title": "SemEval-2017 task 4: Sentiment analysis in Twitter", |
| "authors": [ |
| { |
| "first": "S", |
| "middle": [], |
| "last": "Rosenthal", |
| "suffix": "" |
| }, |
| { |
| "first": "N", |
| "middle": [], |
| "last": "Farra", |
| "suffix": "" |
| }, |
| { |
| "first": "P", |
| "middle": [], |
| "last": "Nakov", |
| "suffix": "" |
| } |
| ], |
| "year": 2017, |
| "venue": "Proceedings of the 11th International Workshop on Semantic Evaluation", |
| "volume": "", |
| "issue": "", |
| "pages": "502--518", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "S. Rosenthal, N. Farra, and P. Nakov. \"SemEval-2017 task 4: Sentiment analysis in Twitter,\" In: Proceedings of the 11th International Workshop on Semantic Evaluation (SemEval-2017), pp. 502-518, 2017.", |
| "links": null |
| }, |
| "BIBREF10": { |
| "ref_id": "b10", |
| "title": "deepSA2018 at SemEval-2018 Task 1: Multi-task Learning of Different Label for Affect in Tweets", |
| "authors": [ |
| { |
| "first": "Z", |
| "middle": [ |
| "Y" |
| ], |
| "last": "Gao", |
| "suffix": "" |
| }, |
| { |
| "first": "C", |
| "middle": [ |
| "P" |
| ], |
| "last": "Chen", |
| "suffix": "" |
| } |
| ], |
| "year": 2018, |
| "venue": "Proceedings of The 12th International Workshop on Semantic Evaluation", |
| "volume": "", |
| "issue": "", |
| "pages": "226--230", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Z.Y. Gao, and C.P. Chen, \"deepSA2018 at SemEval-2018 Task 1: Multi-task Learning of Different Label for Affect in Tweets,\" In: Proceedings of The 12th International Workshop on Semantic Evaluation, pp. 226-230, 2018.", |
| "links": null |
| }, |
| "BIBREF11": { |
| "ref_id": "b11", |
| "title": "Distributed representations of words and phrases and their compositionality", |
| "authors": [ |
| { |
| "first": "T", |
| "middle": [], |
| "last": "Mikolov", |
| "suffix": "" |
| }, |
| { |
| "first": "I", |
| "middle": [], |
| "last": "Sutskever", |
| "suffix": "" |
| }, |
| { |
| "first": "K", |
| "middle": [], |
| "last": "Chen", |
| "suffix": "" |
| }, |
| { |
| "first": "G", |
| "middle": [ |
| "S" |
| ], |
| "last": "Corrado", |
| "suffix": "" |
| } |
| ], |
| "year": 2013, |
| "venue": "Advances in neural information processing systems", |
| "volume": "", |
| "issue": "", |
| "pages": "3111--3119", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "T. Mikolov, I. Sutskever, K. Chen, and G.S. Corrado, \"Distributed representations of words and phrases and their compositionality,\" In: Advances in neural information processing systems, pp. 3111-3119, 2013.", |
| "links": null |
| }, |
| "BIBREF12": { |
| "ref_id": "b12", |
| "title": "Attention-based lstm for aspect-level sentiment classification", |
| "authors": [ |
| { |
| "first": "Y", |
| "middle": [], |
| "last": "Wang", |
| "suffix": "" |
| }, |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Huang", |
| "suffix": "" |
| }, |
| { |
| "first": "L", |
| "middle": [], |
| "last": "Zhao", |
| "suffix": "" |
| } |
| ], |
| "year": 2016, |
| "venue": "Proceedings of the 2016 conference on empirical methods in natural language processing", |
| "volume": "", |
| "issue": "", |
| "pages": "606--615", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Y. Wang, M. Huang, and L. Zhao,\" Attention-based lstm for aspect-level sentiment classification,\" In: Proceedings of the 2016 conference on empirical methods in natural language processing, pp. 606-615, 2016.", |
| "links": null |
| } |
| }, |
| "ref_entries": { |
| "TABREF0": { |
| "text": "Sentiment Analysis, Polarity Classification, Word Vector.", |
| "html": null, |
| "num": null, |
| "type_str": "table", |
| "content": "<table><tr><td colspan=\"3\">0 \u70ba\u4e2d\u6027\u60c5\u7dd2\uff0c\u8cc7\u6599\u5206\u5e03\u60c5\u5f62\u5982\u8868\u4e00\u3002\u70ba\u4e86\u64f4\u5c55\u8a13\u7df4\u8cc7\u6599\uff0c\u6211\u5011\u5148\u5229\u7528 SemEval-2017 [9] \u540d\u7a31 \u8a5e\u5f59\u91cf \u672a\u5c0d\u61c9\u8a5e\u5f59\u91cf \u7684\u4e09\u985e\u8cc7\u6599\u8a13\u7df4\u4e09\u985e\u6a21\u578b\uff0c\u5f97\u5230\u80fd\u6293\u5230\u300c\u53e5\u5b50\u300d\u7684\u4e09\u985e\u60c5\u7dd2\u7279\u5fb5 RNN \u6b0a\u91cd\uff0c\u4e26\u5c07\u9019\u500b \u8868\u4e8c\u3001\u82f1\u6587\u7dad\u57fa\u767e\u79d1\u8207\u81ea\u884c\u6536\u96c6\u63a8\u6587\u8a5e\u5f59\u5c0d\u61c9\u72c0\u6cc1\u4e4b\u6bd4\u8f03 (\u4e8c)\u3001 \u4e03\u985e\u63a8\u6587\u6975\u6027\u5206\u985e\u7cfb\u7d71 ( ) att att M tanh W X b = + (1)</td></tr><tr><td colspan=\"3\">\u6211\u5011\u5206\u985e\u4e03\u985e\u60c5\u7dd2\u5171\u4f7f\u7528\u4e09\u500b\u6a21\u578b\uff0c\u5206\u5225\u662f CNN \u4e03\u985e\u6a21\u578b\u3001RNN \u4e09\u985e\u6a21\u578b\u4ee5\u53ca RNN \u6b0a\u91cd\u505a\u70ba\u63a5\u4e0b\u4f86\u8a13\u7df4\u7684\u4e03\u985e\u6a21\u578b\u7684\u521d\u59cb\u6b0a\u91cd\uff0c\u4f7f\u4e03\u985e\u6a21\u578b\u80fd\u5148\u6293\u51fa\u4e09\u985e\u7684\u7279\u5fb5\u518d\u9032\u800c\u5206 \u6210\u4e03\u985e\uff0c\u4ee5\u63d0\u5347\u6e96\u78ba\u7387\uff0c\u6700\u7d42\u6211\u5011\u4ee5\u4e03\u985e\u6a21\u578b\u7684\u5206\u985e\u7d50\u679c\u70ba\u6700\u5f8c\u7684\u5206\u985e\u7d50\u679c[10]\u3002 \u82f1\u6587\u7dad\u57fa\u767e\u79d1\u8a9e\u6599\u5eab 2,287,131 322 / 4419 \u4e03\u985e\u6a21\u578b\uff0c\u6574\u9ad4\u6a21\u578b\u67b6\u69cb\u5716\u5982\u5716\u4e8c\u3002</td></tr><tr><td colspan=\"3\">\u672c\u6587\u4e3b\u8981\u5206\u70ba\u56db\u500b\u90e8\u5206\uff1a\u7b2c\u4e00\u90e8\u5206\u70ba\u7dd2\u8ad6\uff1b\u7b2c\u4e8c\u90e8\u5206\u70ba\u7814\u7a76\u65b9\u6cd5\uff0c\u4ecb\u7d39\u8cc7\u6599\u7684\u4f7f\u7528\u65b9\u5f0f Tweepy \u81ea\u884c\u6536\u96c6\u63a8\u6587 192,307 163 / 4419</td></tr><tr><td colspan=\"3\">\u8207\u5be6\u9a57\u6d41\u7a0b\uff1b\u7b2c\u4e09\u90e8\u5206\u70ba\u5be6\u9a57\u7d50\u679c\uff0c\u4ecb\u7d39\u5be6\u9a57\u7684\u8a2d\u5b9a\u8207\u7d50\u679c\uff1b\u7b2c\u56db\u90e8\u5206\u70ba\u7d50\u8ad6\u3002</td></tr><tr><td colspan=\"3\">\u8868\u4e00\u3001\u4f7f\u7528\u8cc7\u6599\u5206\u5e03\u60c5\u5f62 \u540c\u6642\uff0c\u6211\u5011\u4e5f\u5c0d\u6536\u96c6\u7684\u63a8\u6587\u524d\u8655\u7406[5]\uff0c\u5305\u542b\u5168\u90e8\u8f49\u70ba\u5c0f\u5beb\u3001\u6a19\u8a18\u5168\u5927\u5beb\u5f37\u8abf\u7684\u8a5e\u6216 hash</td></tr><tr><td colspan=\"3\">\u540d\u7a31 tag \u7684\u8a5e\u3001\u5c07\u4f7f\u7528\u8005\u540d\u7a31\u3001\u7db2\u5740\u7d71\u4e00\u6a19\u8a18\u70ba<USER>\u8207<URL>\uff0c\u4ee5\u6e1b\u4f4e\u8a5e\u5f59\u7684\u8907\u96dc\u5ea6\uff0c \u6b63\u5411 \u4e2d\u6027 \u8ca0\u9762</td></tr><tr><td>\u5716\u4e09\u3001CNN \u4e03\u985e\u6a21\u578b</td><td/><td/></tr><tr><td>1</td><td>0</td><td>-1</td></tr><tr><td colspan=\"3\">\u4e00\u3001\u7dd2\u8ad6 RNN \u4e09\u985e\u6a21\u578b\u5982\u5716\u56db\u6240\u793a\uff0c\u6240\u4f7f\u7528\u7684\u8cc7\u6599\u70ba SemEval-2017 \u7684\u4e09\u985e\u8cc7\u6599\uff0c\u6b64\u6a21\u578b\u7531\u4e00\u5c64 3 2 1 0 -1 -2 -3</td></tr><tr><td colspan=\"3\">\u8fd1\u5e74\u4f86\u6a5f\u5668\u5b78\u7fd2\u84ec\u52c3\u767c\u5c55\uff0c\u5728\u6587\u5b57\u7684\u6975\u6027\u5206\u985e\u4e0a\u4e5f\u6709\u76f8\u7576\u7684\u8ca2\u737b\u3002\u5728\u6587\u5b57\u65b9\u9762\u7684\u4efb\u52d9 SemEval-2017-ALL 9434 16279 7203 \u7684\u96d9\u5411\u9577\u77ed\u671f\u8a18\u61b6(Bidirectional LSTM\uff0cBi-LSTM)\u69cb\u6210\uff0c\u4e3b\u8981\u7684\u76ee\u7684\u70ba\u53d6\u51fa Bi-LSTM</td></tr><tr><td colspan=\"3\">\u4e2d\uff0c\u70ba\u4e86\u8b93\u96fb\u8166\u80fd\u7406\u89e3\u4eba\u985e\u7684\u6587\u5b57\uff0c\u6211\u5011\u9700\u8981\u5c07\u6bcf\u500b\u8a5e\u8f49\u63db\u6210\u8a5e\u5411\u91cf\uff0c\u82e5\u5169\u500b\u8a5e\u7684\u8a5e\u5411 SemEval-2018 \u8a13\u7df4\u96c6 125 92 167 341 78 249 129 \u7684\u6b0a\u91cd\uff0c\u4e26\u7528\u65bc\u6700\u5f8c RNN \u4e03\u985e\u6a21\u578b\u4e2d\u521d\u59cb\u5316\u7b2c\u4e00\u5c64 Bi-LSTM \u7684\u6b0a\u91cd\uff0c\u5982\u5716\u4e8c\u7da0\u8272\u90e8\u5206</td></tr><tr><td colspan=\"3\">\u91cf\u8ddd\u96e2\u6108\u8fd1\u5247\u8868\u793a\u5169\u500b\u8a5e\u7684\u610f\u7fa9\u76f8\u8fd1\uff0c\u6108\u9060\u5247\u8868\u793a\u610f\u601d\u76f8\u5dee\u6108\u9060\uff0cWord2vec [1]\u3001GloVe SemEval-2018 \u9a57\u8b49\u96c6 53 35 58 105 34 95 69 \u6240\u793a\uff0c\u4f7f\u4e03\u985e\u6a21\u578b\u80fd\u5148\u5c07\u63a8\u6587\u53e5\u5b50\u4e2d\u627e\u51fa\u4e09\u985e\u7684\u7279\u5fb5\uff0c\u518d\u85c9\u7531\u7b2c\u4e8c\u5c64\u7684 Bi-LSTM \u9032\u4e00</td></tr><tr><td colspan=\"3\">[2]\u3001fastText [3]\u90fd\u662f\u5e38\u898b\u8a13\u7df4\u8a5e\u5411\u91cf\u7684\u65b9\u6cd5\uff0c\u800c\u8a13\u7df4\u51fa\u4f86\u8a5e\u5411\u91cf\u7684\u512a\u52a3\u6703\u5927\u5927\u7684\u5f71\u97ff\u6975 SemEval-2018 \u6e2c\u8a66\u96c6 137 91 107 262 80 167 93 \u5716\u4e8c\u3001\u63a8\u6587\u6975\u6027\u5206\u985e\u6574\u9ad4\u67b6\u69cb\u5716 \u6b65\u5206\u6210\u4e03\u985e\u7684\u60c5\u7dd2\u3002</td></tr><tr><td colspan=\"3\">\u6027\u5206\u985e\u7684\u7d50\u679c\u3002\u5728\u5206\u985e\u6a21\u578b\u65b9\u9762\uff0c\u7121\u8ad6\u662f CNN \u6a21\u578b [4]\u3001RNN \u6a21\u578b [5]\uff0c\u5747\u53d6\u5f97\u6bd4\u50b3</td></tr><tr><td colspan=\"3\">\u7d71\u57fa\u65bc\u7d71\u8a08\u7684\u65b9\u6cd5\u66f4\u597d\u7684\u6210\u679c\u3002\u5728 CNN \u6a21\u578b\u4e2d\uff0c\u4f7f\u7528\u4e0d\u540c\u5927\u5c0f\u7684\u5377\u7a4d\u6838(Kernel)\u4ee3 \u4e8c\u3001\u7814\u7a76\u65b9\u6cd5 CNN \u4e03\u985e\u6a21\u578b\u5982\u5716\u4e09\u6240\u793a\uff0c\u5728\u9019\u500b\u6a21\u578b\u4e2d\uff0c\u6211\u5011\u4f7f\u7528 SemEval-2018 Task1: Affect in</td></tr><tr><td colspan=\"3\">\u8868\u4e00\u6b21\u8b93\u795e\u7d93\u7db2\u8def\u770b\u591a\u5c11\u500b\u8a5e\u4e26\u53d6\u51fa\u5b83\u5011\u7684\u60c5\u7dd2\u7279\u5fb5\uff0c\u82e5 \u5377\u7a4d\u6838\u5927\u5c0f\u70ba 1 \u5247\u4ee3\u8868\u4e00\u6b21\u770b Tweets \u5b50\u4efb\u52d9 V-oc \u4e4b\u4e03\u985e\u8a13\u7df4\u8cc7\u6599\uff0c\u4f46\u9019\u500b\u6a21\u578b\u4e26\u975e\u5f97\u5230\u6700\u7d42\u5206\u985e\u7d50\u679c\u7684\u6a21\u578b\uff0c\u800c\u662f</td></tr><tr><td colspan=\"3\">\u4e00\u500b\u8a5e\uff0c\u82e5\u70ba 2 \u5247\u4e00\u6b21\u770b\u5169\u500b\u8a5e\uff0c\u4e26\u900f\u904e\u6700\u5927\u6c60\u5316\u627e\u51fa\u5c0d\u60c5\u7dd2\u5f71\u97ff\u6700\u5927\u7684\u8a5e\u6216\u8a5e\u7d44\uff0c\u6700 (\u4e00) \u3001\u9810\u8a13\u7df4\u8a5e\u5411\u91cf \u85c9\u7531\u6b64\u6a21\u578b\u4f86\u5c0d\u5d4c\u5165\u5c64(Embedding Layer)\u4e2d\u7684\u8a5e\u5411\u91cf\u9032\u884c\u5fae\u8abf\uff0c\u7528\u53cd\u5411\u50b3\u64ad\u81f3\u5d4c\u5165</td></tr><tr><td colspan=\"3\">\u5f8c\u5f97\u51fa\u5206\u985e\u7d50\u679c\u3002\u800c\u5728 RNN \u6a21\u578b\u4e2d\uff0c\u5247\u5e38\u4f7f\u7528\u9577\u77ed\u671f\u8a18\u61b6(Long Short-Term Memory\uff0c Twitter \u4e0a\u7684\u63a8\u6587\u4e0d\u50c5\u6709\u5b57\u6578\u7684\u9650\u5236\uff0c\u63a8\u6587\u901a\u5e38\u90fd\u662f\u77ed\u77ed\u7684\u53e5\u5b50\uff0c\u53e5\u5b50\u5167\u5bb9\u5305\u542b\u5404\u7a2e emoji \u5c64\u7684\u65b9\u5f0f\uff0c\u4f7f\u6b63\u9762\u60c5\u7dd2\u8207\u8ca0\u9762\u60c5\u7dd2\u7684\u8a5e\u5411\u91cf\u4e0d\u518d\u90a3\u9ebc\u63a5\u8fd1\uff0c\u4e26\u5c07\u9019\u4efd\u8abf\u6574\u904e\u7684\u8a5e\u5411\u91cf\u7528</td></tr><tr><td colspan=\"3\">LSTM)[6]\u4f86\u8655\u7406\uff0c\u6574\u5408\u6574\u53e5\u53e5\u5b50\u7684\u6db5\u7fa9\u7136\u5f8c\u505a\u51fa\u5224\u65b7\u3002\u5169\u7a2e\u6a21\u578b\u67b6\u69cb\u6240\u6ce8\u91cd\u7684\u91cd\u9ede\u4e0d \u8207\u8868\u60c5\u7b26\u865f\uff0c\u6587\u6cd5\u4e5f\u4e0d\u5982\u5176\u4ed6\u8a9e\u6599\u5eab\u62d8\u8b39\uff0c\u751a\u81f3\u5e38\u6703\u6709\u5404\u5f0f\u82f1\u6587\u7e2e\u5beb\u51fa\u73fe\uff0c\u5982\uff1aOMG \u5229\u65bc\u60c5\u7dd2\u5206\u6790\u7684\u4efb\u52d9\uff0c\u5728\u4e0b\u4e00\u5c0f\u7bc0\u4e2d\uff0c\u6211\u5011\u5c07\u6703\u6e1b\u7de9\u9019\u500b\u554f\u984c\u3002 \u65bc\u63a5\u4e0b\u4f86 RNN \u6a21\u578b\u7684\u5d4c\u5165\u5c64\u4e0a\uff0c\u5982\u5716\u4e8c\u4e2d\u9ec3\u8272\u90e8\u5206\u6240\u793a\uff0c\u4ee5\u5e6b\u52a9\u6211\u5011\u5728\u4e4b\u5f8c\u7684\u8a13\u7df4\u904e</td></tr><tr><td colspan=\"3\">\u540c\uff0cCNN \u504f\u5411\u627e\u51fa\u5f71\u97ff\u6700\u5927\u7684\u8a5e\u6216\u8a5e\u7d44\uff0c\u800c RNN \u5247\u662f\u91dd\u5c0d\u6574\u53e5\u53e5\u5b50\u5f97\u51fa\u7d50\u679c\u3002 (Oh My God)\u7b49\u3002\u56e0\u6b64\uff0c\u82e5\u662f\u4f7f\u7528\u4e00\u822c\u8a9e\u6599\u5eab\u8a13\u7df4\u8a5e\u5411\u91cf\u6703\u6709\u5927\u91cf\u7684\u8868\u60c5\u7b26\u865f\u6216\u7db2\u8def \u7a0b\u4e2d\uff0c\u66f4\u597d\u7684\u9032\u884c\u60c5\u7dd2\u7684\u5206\u985e\u3002 \u5728\u6211\u5011\u7684\u6a21\u578b\u7cfb\u7d71\u4e2d\uff0c\u6211\u5011\u7d50\u5408 CNN \u8207 RNN \u7684\u512a\u9ede\uff0c\u5229\u7528 CNN \u504f\u91cd\u65bc\u300c\u8a5e\u300d\u7684\u7279\u6027\uff0c \u5728\u8a13\u7df4\u60c5\u7dd2\u5206\u6790\u7684\u795e\u7d93\u7db2\u8def\u6642\u5141\u8a31\u53cd\u5411\u50b3\u64ad(Back Propagation)\u81f3\u5d4c\u5165\u5c64\u7684\u8a5e\u5411\u91cf\uff0c \u4e0a\u6d41\u884c\u7684\u7e2e\u5beb\u7121\u6cd5\u5305\u542b\u5176\u4e2d\uff0c\u800c\u9019\u4e9b\u90e8\u5206\u5f80\u5f80\u90fd\u662f\u60c5\u7dd2\u6700\u5f37\u70c8\u7684\u5730\u65b9\uff0c\u6240\u4ee5\u6211\u5011\u63a1\u7528\u81ea \u5728\u6b64\u6a21\u578b\uff0c\u5728\u8a5e\u5d4c\u5165\u5f8c\u7d93\u904e\u5927\u5c0f\u5206\u5225\u70ba 1\u30012\u30013 \u7684\u5377\u7a4d\u6838\uff0c\u76f8\u7576\u65bc\u4ee5\u4e00\u5143\u8a9e\u6cd5 (Unigram)\u3001</td></tr><tr><td colspan=\"3\">\u4f7f\u8a5e\u5411\u91cf\u5e36\u6709\u60c5\u7dd2\u7279\u5fb5[7]\uff0c\u4f86\u89e3\u6c7a\u8a5e\u5411\u91cf\u53ef\u80fd\u6b63\u8ca0\u9762\u60c5\u7dd2\u8a5e\u4e4b\u9593\u8ddd\u96e2\u904e\u65bc\u63a5\u8fd1\uff0c\u800c\u4f7f \u884c\u5f9e Twitter \u6536\u96c6\u63a8\u6587\u7684\u65b9\u5f0f\uff0c\u85c9\u7531 Twitter \u5b98\u65b9\u63d0\u4f9b\u7684 Tweepy \u9019\u9805\u5de5\u5177\u4f86\u53d6\u5f97\u63a8\u6587\uff0c \u4e8c\u5143\u8a9e\u6cd5(Bigram) \u3001\u4e09\u5143\u8a9e\u6cd5(Trigram)\u7684\u89d2\u5ea6\u4f86\u770b\u63a8\u6587\uff0c\u4e26\u5404\u81ea\u53d6 Global max pooling \u5716\u56db\u3001RNN \u4e09\u985e\u6a21\u578b</td></tr><tr><td colspan=\"3\">\u795e\u7d93\u7db2\u8def\u7121\u6cd5\u5f97\u77e5\u8a72\u8a5e\u70ba\u6b63\u9762\u60c5\u611f\u6216\u8ca0\u9762\u60c5\u611f\u7684\u554f\u984c\u3002\u63a5\u8457\u6211\u5011\u4f7f\u7528\u8abf\u6574\u5f8c\u7684\u8a5e\u5411\u91cf\u65bc \u6536\u96c6\u671f\u9593\u70ba 2/27 \u81f3 4/13\uff0c\u63a8\u6587\u6578\u7d04 22,000,000 \u5247\uff0c\u8a5e\u5f59\u91cf\u7d04 192,000 \u500b\u8a5e\uff0c\u8868\u4e8c\u70ba\u82f1\u6587 \u627e\u51fa\u5f71\u97ff\u6700\u5927\u7684 n-gram \u4e26\u4e32\u63a5\u5728\u4e00\u8d77\uff0c\u6700\u5f8c\u5f97\u51fa\u5206\u985e\u7d50\u679c\uff0c\u4e14\u5141\u8a31\u53cd\u5411\u50b3\u64ad\u81f3\u5d4c\u5165\u5c64\uff0c RNN \u4e03\u985e\u6a21\u578b\u5982\u5716\u4e94\u6240\u793a\uff0c\u4f7f\u7528\u8cc7\u6599\u70ba SemEval-2018 \u7684\u4e03\u985e\u8cc7\u6599\uff0c\u4e5f\u662f\u5f97\u51fa\u6700\u5f8c\u6975\u6027</td></tr><tr><td colspan=\"3\">\u504f\u91cd\u300c\u53e5\u5b50\u300d\u7684 RNN \u6a21\u578b\u4e2d\u8a13\u7df4\u3002\u5728 RNN \u6a21\u578b\u4e2d\uff0c\u7531\u65bc\u5728\u6211\u5011\u4e3b\u8981\u505a\u70ba\u8a55\u4f30\u6a19\u6e96\u7684 \u7dad\u57fa\u767e\u79d1\u8a9e\u6599\u5eab\u8207\u81ea\u884c\u6536\u96c6\u4e4b\u63a8\u6587\u8cc7\u6599\u96c6\u5728 SemEval-2018 \u8a13\u7df4\u96c6\u4e0a\u662f\u5426\u627e\u5230\u5c0d\u61c9\u8a5e\u5f59 \u8abf\u6574\u8a5e\u5411\u91cf\uff0c\u4e26\u5c07\u9019\u4efd\u8abf\u6574\u904e\u5f8c\u7684\u8a5e\u5411\u91cf\u7d66\u5f8c\u7e8c\u7684\u6a21\u578b\u4f7f\u7528\u3002 \u5206\u985e\u7d50\u679c\u7684\u6a21\u578b\uff0c\u7531\u5169\u5c64\u7684 Bi-LSTM \u6240\u7d44\u6210\uff0c\u7b2c\u4e00\u5c64 Bi-LSTM \u7531 RNN \u4e09\u985e\u6a21\u578b\u8a13\u7df4</td></tr><tr><td colspan=\"3\">SemEval-2018 Task1: Affect in Tweets \u5b50\u4efb\u52d9 V-oc \u70ba\u4e03\u985e +3 \u5230 -3 \u7684\u60c5\u7dd2\u5206\u985e\u554f\u984c \u7684\u72c0\u6cc1\uff0c\u53ef\u4ee5\u767c\u73fe\u5230\u81ea\u884c\u6536\u96c6\u7684\u63a8\u6587\u8cc7\u6599\u96c6\u5728 Twitter \u9019\u500b\u6587\u6cd5\u4e0d\u62d8\u8b39\u4e14\u6709\u975e\u5e38\u591a\u65b0\u5275 \u5f97\u51fa\u7684\u6b0a\u91cd\u521d\u59cb\u5316\uff0c\u4e14\u63a1\u7528\u6ce8\u610f\u529b\u6a5f\u5236(Attention)[12]\uff0c\u4f7f\u53e5\u5b50\u4e2d\u91cd\u8981\u7684\u5730\u65b9\u6709\u8457\u8f03</td></tr><tr><td colspan=\"3\">[8]\uff0c+3 \u8868\u793a\u6700\u6b63\u9762\u7684\u60c5\u7dd2\u3001-3 \u8868\u793a\u6700\u8ca0\u9762\u7684\u60c5\u7dd2\u30010 \u5247\u662f\u4e2d\u6027\u60c5\u7dd2\uff0c\u800c SemEval-2017 \u8a5e\u7684\u9818\u57df\u4e0a\uff0c\u8868\u73fe\u512a\u65bc\u8a5e\u5f59\u91cf\u70ba\u5341\u500d\u7684\u82f1\u6587\u7dad\u57fa\u8a9e\u6599\u5eab\uff0c\u6709\u66f4\u597d\u7684\u5c0d\u61c9\u95dc\u4fc2\u3002 \u9ad8\u5f71\u97ff\u6700\u5f8c\u6975\u6027\u5206\u985e\u7684\u80fd\u529b\uff0c\u516c\u5f0f\u5982\u4e0b\uff1a</td></tr><tr><td colspan=\"3\">\u8207\u4e4b\u524d\u7684\u8cc7\u6599\u7686\u70ba\u4e09\u985e\u60c5\u7dd2\u5206\u5225\u70ba -1\u30010\u3001+1 \u7684\u8cc7\u6599\uff0c+1 \u70ba\u6b63\u9762\u60c5\u7dd2\u3001-1 \u70ba\u8ca0\u9762\u60c5\u7dd2\u3001 \u5716\u4e00\u3001Skip-gram \u6a21\u578b\u67b6\u69cb</td></tr></table>" |
| } |
| } |
| } |
| } |
