| { | |
| "paper_id": "2019", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T14:55:05.304782Z" | |
| }, | |
| "title": "Stance Detection of Social Network Users by combining Latent Dirichlet Allocation and Support Vector Machine", | |
| "authors": [ | |
| { | |
| "first": "I-Huan", | |
| "middle": [], | |
| "last": "\u7fc1\u7fca\u6853", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "National Taipei University of Technology", | |
| "location": {} | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "\u570b\u7acb\u81fa\u5317\u79d1\u6280\u5927\u5b78\u8cc7\u8a0a\u5de5\u7a0b\u5b78\u7cfb", | |
| "middle": [], | |
| "last": "Weng", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "National Taipei University of Technology", | |
| "location": {} | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "Jenq-Haur", | |
| "middle": [], | |
| "last": "\u738b\u6b63\u8c6a", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "Taipei University of Technology", | |
| "location": {} | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "\u570b\u7acb\u81fa\u5317\u79d1\u6280\u5927\u5b78\u8cc7\u8a0a\u5de5\u7a0b\u5b78\u7cfb", | |
| "middle": [], | |
| "last": "Wang", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "Taipei University of Technology", | |
| "location": {} | |
| }, | |
| "email": "[email protected]" | |
| } | |
| ], | |
| "year": "", | |
| "venue": null, | |
| "identifiers": {}, | |
| "abstract": "", | |
| "pdf_parse": { | |
| "paper_id": "2019", | |
| "_pdf_hash": "", | |
| "abstract": [], | |
| "body_text": [ | |
| { | |
| "text": "In traditional stance analysis, questionnaire survey or telephone survey are often used to know the opinions of each person under different topics. However, due to the traditional statistical methods, the sample size is too small to get good result. Existing methods are usually based on sentiment lexicon, Convolutional Neural Network (CNN) or Recurrent Neural Network (RNN). And the text features are based on N-Gram or TF-IDF, which do not help to understand the semantics of the text. This research proposes to use Word2Vec for word embedding and combine the LDA to obtain the text feature. For stance detection, we use Support Vector Machine (SVM) to train the classifier to detect the subjectivity of texts, and to predict user stances.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Abstract", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "In the experiment, we used the data from SemEval-2016 Stance Detection Task, and use a variety of evaluation methods (F-Measure, Accuracy, Precision, Recall) to evaluate performance. Compared with SemEval-2016 official baseline and other teams scores, our proposed method can get better result on average (F-Measure : 83.36%). ", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Abstract", | |
| "sec_num": null | |
| } | |
| ], | |
| "back_matter": [], | |
| "bib_entries": { | |
| "BIBREF0": { | |
| "ref_id": "b0", | |
| "title": "Takelab at semeval-2016 task 6: stance classification in tweets using a genetic algorithm based ensemble", | |
| "authors": [], | |
| "year": 2016, | |
| "venue": "Proceedings of the 10th international workshop on semantic evaluation", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "464--468", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "2016, June). Takelab at semeval-2016 task 6: stance classification in tweets using a genetic algorithm based ensemble. In Proceedings of the 10th international workshop on semantic evaluation (SemEval-2016)(pp. 464-468).", | |
| "links": null | |
| }, | |
| "BIBREF1": { | |
| "ref_id": "b1", | |
| "title": "Idi@ ntnu at semeval-2016 task 6: Detecting stance in tweets using shallow features and glove vectors for word representation", | |
| "authors": [ | |
| { | |
| "first": "H", | |
| "middle": [], | |
| "last": "B\u00f8hler", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "P", | |
| "middle": [], | |
| "last": "Asla", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "E", | |
| "middle": [], | |
| "last": "Marsi", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "R", | |
| "middle": [], | |
| "last": "Sae Tre", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2016, | |
| "venue": "Proceedings of the 10th International Workshop on Semantic Evaluation", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "445--450", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "B\u00f8hler, H., Asla, P., Marsi, E., & Sae tre, R. (2016, June). Idi@ ntnu at semeval-2016 task 6: Detecting stance in tweets using shallow features and glove vectors for word representation. In Proceedings of the 10th International Workshop on Semantic Evaluation (SemEval-2016) (pp. 445-450).", | |
| "links": null | |
| }, | |
| "BIBREF2": { | |
| "ref_id": "b2", | |
| "title": "pkudblab at semeval-2016 task 6: A specific convolutional neural network system for effective stance detection", | |
| "authors": [ | |
| { | |
| "first": "W", | |
| "middle": [], | |
| "last": "Wei", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "X", | |
| "middle": [], | |
| "last": "Zhang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "X", | |
| "middle": [], | |
| "last": "Liu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "W", | |
| "middle": [], | |
| "last": "Chen", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "T", | |
| "middle": [], | |
| "last": "Wang", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2016, | |
| "venue": "Proceedings of the 10th international workshop on semantic evaluation", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "384--388", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Wei, W., Zhang, X., Liu, X., Chen, W., & Wang, T. (2016, June). pkudblab at semeval-2016 task 6: A specific convolutional neural network system for effective stance detection. In Proceedings of the 10th international workshop on semantic evaluation (SemEval-2016) (pp. 384-388).", | |
| "links": null | |
| }, | |
| "BIBREF3": { | |
| "ref_id": "b3", | |
| "title": "DeepStance at SemEval-2016 Task 6: Detecting Stance in Tweets Using Character and Word-Level CNNs", | |
| "authors": [ | |
| { | |
| "first": "P", | |
| "middle": [], | |
| "last": "Vijayaraghavan", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "I", | |
| "middle": [], | |
| "last": "Sysoev", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "S", | |
| "middle": [], | |
| "last": "Vosoughi", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "D", | |
| "middle": [], | |
| "last": "Roy", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2016, | |
| "venue": "Proceedings of the 10th International Workshop on Semantic Evaluation", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "413--419", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Vijayaraghavan, P., Sysoev, I., Vosoughi, S., & Roy, D. (2016, June). DeepStance at SemEval-2016 Task 6: Detecting Stance in Tweets Using Character and Word-Level CNNs. In Proceedings of the 10th International Workshop on Semantic Evaluation (SemEval-2016) (pp. 413-419).", | |
| "links": null | |
| }, | |
| "BIBREF4": { | |
| "ref_id": "b4", | |
| "title": "Mitre at semeval-2016 task 6: Transfer learning for stance detection", | |
| "authors": [ | |
| { | |
| "first": "G", | |
| "middle": [], | |
| "last": "Zarrella", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [ | |
| "J" | |
| ], | |
| "last": "", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2016, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Zarrella, G. and A.J.a.p.a. Marsh, Mitre at semeval-2016 task 6: Transfer learning for stance detection. 2016.", | |
| "links": null | |
| }, | |
| "BIBREF5": { | |
| "ref_id": "b5", | |
| "title": "Tohoku at SemEval-2016 task 6: feature-based model versus convolutional neural network for stance detection", | |
| "authors": [ | |
| { | |
| "first": "Y", | |
| "middle": [], | |
| "last": "Igarashi", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "H", | |
| "middle": [], | |
| "last": "Komatsu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "S", | |
| "middle": [], | |
| "last": "Kobayashi", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "N", | |
| "middle": [], | |
| "last": "Okazaki", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "K", | |
| "middle": [], | |
| "last": "Inui", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2016, | |
| "venue": "Proceedings of the 10th International Workshop on Semantic Evaluation", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "401--407", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Igarashi, Y., Komatsu, H., Kobayashi, S., Okazaki, N., & Inui, K. (2016, June). Tohoku at SemEval-2016 task 6: feature-based model versus convolutional neural network for stance detection. In Proceedings of the 10th International Workshop on Semantic Evaluation (SemEval-2016) (pp. 401-407).", | |
| "links": null | |
| }, | |
| "BIBREF7": { | |
| "ref_id": "b7", | |
| "title": "Semeval-2016 task 6: Detecting stance in tweets", | |
| "authors": [], | |
| "year": 2016, | |
| "venue": "Proceedings of the 10th International Workshop on Semantic Evaluation", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "31--41", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Semeval-2016 task 6: Detecting stance in tweets. In Proceedings of the 10th International Workshop on Semantic Evaluation (SemEval-2016) (pp. 31-41).", | |
| "links": null | |
| }, | |
| "BIBREF8": { | |
| "ref_id": "b8", | |
| "title": "Latent dirichlet allocation", | |
| "authors": [ | |
| { | |
| "first": "D", | |
| "middle": [ | |
| "M" | |
| ], | |
| "last": "Blei", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [ | |
| "Y" | |
| ], | |
| "last": "Ng", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [ | |
| "I J J O M L R" | |
| ], | |
| "last": "Jordan", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2003, | |
| "venue": "", | |
| "volume": "3", | |
| "issue": "", | |
| "pages": "993--1022", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Blei, D.M., A.Y. Ng, and M.I.J.J.o.m.L.r. Jordan, Latent dirichlet allocation. 2003. 3(Jan): p. 993-1022.", | |
| "links": null | |
| }, | |
| "BIBREF9": { | |
| "ref_id": "b9", | |
| "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": { | |
| "arXiv": [ | |
| "arXiv:1301.3781" | |
| ] | |
| }, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Mikolov, T., Chen, K., Corrado, G., & Dean, J. (2013). Efficient estimation of word representations in vector space. arXiv preprint arXiv:1301.3781.", | |
| "links": null | |
| } | |
| }, | |
| "ref_entries": { | |
| "FIGREF0": { | |
| "uris": null, | |
| "num": null, | |
| "type_str": "figure", | |
| "text": "Statista, Number of social media users worldwide from 2010 to 2020,https://www.statista.com/statistics/278414/number-of-worldwide-social-network-us ers/ (Viewed on 2019/07/02) [2] Jannati, R., Mahendra, R., Wardhana, C. W., & Adriani, M. (2018, November). Stance Classification Towards Political Figures on Blog Writing. In 2018 International Conference on Asian Language Processing (IALP) (pp. 96-101). [3] Tumasjan, A., Sprenger, T. O., Sandner, P. G., & Welpe, I. M. (2010, May). Predicting elections with twitter: What 140 characters reveal about political sentiment. In Fourth international AAAI conference on weblogs and social media." | |
| }, | |
| "FIGREF1": { | |
| "uris": null, | |
| "num": null, | |
| "type_str": "figure", | |
| "text": "Sasaki, A., Mizuno, J., Okazaki, N., & Inui, K. (2016, October). Stance classification by recognizing related events about targets. In 2016 IEEE/WIC/ACM International Conference on Web Intelligence (WI) (pp. 582-587). IEEE. [5] Tutek, M., Sekulic, I., Gombar, P., Paljak, I., Culinovic, F., Boltuzic, F., ... & \u0160najder, J." | |
| }, | |
| "TABREF0": { | |
| "text": "Tutek \u7b49\u4eba[5]\u900f\u904e SVM \u5c0d\u6587\u672c\u4f5c N-Gram \u4f86\u9032\u884c\u7acb\u5834\u6aa2\u6e2c\uff1bB\u00f8hler \u7b49 \u4eba[6]\u4f7f\u7528 GloVe \u8a5e\u5411\u91cf\u6a21\u578b\uff0c\u6bd4\u8f03\u4e86 Naive Bayes, SVM \u9019\u5169\u7a2e\u5206\u985e\u6f14\u7b97\u6cd5\u7684\u6548\u679c\uff0c\u767c \u73fe\u7d50\u5408 GloVe \u8a5e\u5411\u91cf\u7684\u7279\u5fb5\u80fd\u63d0\u9ad8\u6aa2\u6e2c\u6548\u679c\uff1b[7, 8]\u4f7f\u7528\u5377\u7a4d\u795e\u7d93\u7db2\u8def(CNN)\u4f86\u9032\u884c\u7acb\u5834 \u6aa2\u6e2c\uff0c\u5176\u4e2d\u4f7f\u7528 Semeval-2016 \u7684\u8cc7\u6599\u96c6\u5728 F-measure \u6709 67.33%\uff1bZarrella \u7b49\u4eba[9]\u900f\u904e\u905e \u6b78\u795e\u7d93\u7db2\u8def(RNN)\u4e26\u4f7f\u7528\u9810\u8a13\u7df4\u7684\u7279\u5fb5\u4f86\u9032\u884c\u7acb\u5834\u6aa2\u6e2c\uff0c\u4f7f\u7528 Semeval-2016 \u7684\u8cc7\u6599\u96c6\u5728 F-measure \u6709 67.8%\u3002\u6839\u64da Igarashi \u7b49\u4eba[10]\u7684\u89c0\u5bdf\u767c\u73fe\uff0c\u6df1\u5ea6\u5b78\u7fd2\u65b9\u9762\u6548\u679c\u6c92\u6709\u4e00\u822c\u7684 \u5206\u985e\u597d\uff1b\u800c Mohammad \u7b49\u4eba[11]\u4e5f\u767c\u73fe\u5728\u7acb\u5834\u6aa2\u6e2c\u4e0a\uff0cSVM \u7684\u6574\u9ad4\u5e73\u5747\u9ad8\u65bc\u5176\u4ed6\u6a21\u578b\u3002 \u57fa\u65bc\u8cc7\u6599\u96c6\u7684\u6578\u91cf\u8207\u524d\u8005\u7684\u89c0\u5bdf\uff0c\u6211\u5011\u63a1\u7528 SVM \u4f5c\u70ba\u4e3b\u8981\u7684\u65b9\u6cd5\u9032\u884c\u7acb\u5834\u6aa2\u6e2c\uff0c\u4e26\u5728 \u5be6\u9a57\u4e2d\u8207 Naive Bayes, LSTM \u7b49\u5206\u985e\u5668\u4f5c\u6bd4\u8f03\u3002 \u5728\u76f8\u95dc\u7684\u8ad6\u6587\u4e2d\uff0c\u5927\u90e8\u5206\u7684\u5e73\u53f0\u90fd\u662f\u4ee5 Facebook \u6216\u662f Twitter \u5e73\u53f0\u70ba\u4e3b\u4f5c\u5206\u6790\uff0c\u4e14\u4e3b\u984c \u5927\u591a\u90fd\u570d\u7e5e\u5728\u653f\u6cbb\u4e0a\u3002\u800c\u76ee\u524d\u5e38\u7528\u7684\u65b9\u6cd5\u662f\u4f7f\u7528\u91dd\u5c0d\u6587\u672c\u4e2d\u76ee\u6a19\u7684\u60c5\u7dd2\u6975\u6027\uff0c\u9019\u6a23\u65b9\u6cd5 \u6709\u4e9b\u7f3a\u9ede\uff0c\u50cf\u5728\u55ae\u4e00\u76ee\u6a19\u6216\u8de8\u9818\u57df\u6642\u6548\u679c\u901a\u5e38\u4e0d\u4f73\uff0c\u70ba\u4e86\u63d0\u5347\u7acb\u5834\u7684\u4e3b\u984c\u591a\u6a23\u6027\u548c\u6e96\u78ba \u7387\uff0c\u672c\u8ad6\u6587\u5c07\u5229\u7528 LDA \u4e3b\u984c\u6a21\u578b\u7684\u7279\u6027\uff0c\u7d50\u5408 LDA \u7acb\u5834\u6a21\u578b\u8207 Word embedding\uff0c\u671f \u662f\u7531 Google \u7684 Tomas Mikolov \u7b49\u4eba\u65bc 2013 \u5e74\u6240\u63d0\u51fa\u7684\u4e00\u7a2e Word", | |
| "content": "<table><tr><td>\u95dc\u9375\u8a5e\uff1a\u7acb\u5834\u6aa2\u6e2c\u3001\u6a5f\u5668\u5b78\u7fd2\u3001\u793e\u7fa4\u7db2\u8def\u5206\u6790\u3001Word2vec\u3001\u96b1\u542b\u72c4\u5229\u514b\u96f7\u5206\u5e03 Keywords: Stance Detection, Machine learning , Social network analysis , Word2vec \u4e00\u3001\u7dd2\u8ad6 \u4eba\u5011\u5728\u751f\u6d3b\u4e2d\u91dd\u5c0d\u7684\u76ee\u6a19\u4e0d\u540c\u6642\uff0c\u4eba\u5011\u7684\u7acb\u5834\u4e5f\u6703\u8b8a\u5f97\u4e0d\u4e00\u6a23\uff0c\u901a\u5e38\u4eba\u5011\u6709\u6240\u885d\u7a81\u6642\u5927 \u5316\u4f86\u8655\u7406\u6578\u64da\u7684\u5206\u6790\u5c07\u662f\u672a\u4f86\u7684\u8da8\u52e2\u3002 \u7531\u65bc\u73fe\u6709\u6587\u672c\u8655\u7406\u901a\u5e38\u90fd\u53ea\u4ee5\u63d0\u7528\u8a5e\u79fb\u9664(Stopword removal)\u4f5c\u524d\u7f6e\u8655\u7406\uff0c\u56e0\u6b64\u672c\u8ad6\u6587\u53e6 \u5916\u518d\u4f7f\u7528\u8a5e\u578b\u9084\u539f(Lemmatization)\u8207\u8a5e\u5e79\u63d0\u53d6(Stemming)\u7684\u65b9\u6cd5\u4f86\u8a55\u4f30\u5c0d\u65bc\u5206\u985e\u5668\u7684\u5f71 \u97ff\u3002\u5728\u7acb\u5834\u6aa2\u6e2c\u4e0a\uff0c\u672c\u8ad6\u6587\u63d0\u51fa\u4ee5\u900f\u904e LDA \u7684\u65b9\u6cd5\uff0c\u4f86\u7522\u751f\u4e3b\u984c\u7279\u5fb5\u4e26\u8207\u7d93\u7531 Word2Vec \u6240\u8f49\u63db\u7684\u7279\u5fb5\u5411\u91cf\u4f5c\u7d50\u5408\uff0c\u4f86\u9032\u884c\u7acb\u5834\u7684\u6aa2\u6e2c\u3002 \u6700\u5f8c\uff0c\u6839\u64da\u672c\u8ad6\u6587\u4e4b\u5be6\u9a57\uff0c\u5728\u4f7f\u7528 LDA \u7d50\u5408\u7279\u5fb5\u5411\u91cf\u6642\uff0c\u78ba\u5be6\u80fd\u4f7f\u5404\u500b\u76ee\u6a19\u4e3b\u984c\u7684\u6e96 \u78ba\u7387\u63d0\u5347\uff0c\u6700\u9ad8\u7684 F-Measure \u70ba 84.23%\uff0c\u5e73\u5747 F-Measure \u70ba 76.88%\u7686\u9ad8\u65bc SemEval-2016 \u4e0a\u7684 Baselines \u8207\u5176\u4ed6\u968a\u4f0d\uff0c\u56e0\u6b64\u53ef\u4ee5\u9a57\u8b49\u6240\u63d0\u51fa\u65b9\u6cd5\u662f\u6709\u6548\u7684\u3002 \u4e8c\u3001\u76f8\u95dc\u7814\u7a76 Jannati \u7b49\u4eba[2]\u900f\u904e\u6536\u96c6\u90e8\u843d\u683c\u7684\u6587\u7ae0\u4f86\u6aa2\u6e2c\u5c0d\u65bc\u653f\u6cbb\u4eba\u7269\u7684\u7acb\u5834\uff0c\u4e26\u4ee5\u60c5\u7dd2\u8fad\u5178\u4f5c\u5206 \u6790\uff1bTumasjan \u7b49\u4eba[3]\u5c0d Twitter \u4e0a\u7684\u8cbc\u6587\u9032\u884c\u5206\u6790\u4e26\u4f9d\u5176\u7d50\u679c\u4f86\u9810\u6e2c 2009 \u5e74\u7684\u5fb7\u570b\u7e3d \u7406\u5927\u9078\uff0c\u4f86\u8b49\u5be6\u793e\u7fa4\u5e73\u53f0\u4e0a\u7684\u8cbc\u6587\u78ba\u5be6\u53ef\u4ee5\u53cd\u6620\u51fa\u4eba\u5011\u7684\u7acb\u5834\uff1bSasaki[4]\u7b49\u4eba\u5247\u662f\u5c0d Twitter \u7684\u8cbc\u6587\u6dfb\u52a0\u984d\u5916\u7684\u63d0\u793a\u6a19\u7c64\uff0c\u5be6\u9a57\u767c\u73fe\u900f\u904e\u6dfb\u52a0\u7684\u984d\u5916\u4e8b\u4ef6\u7279\u5fb5\u53ef\u4ee5\u63d0\u5347\u7acb\u5834 \u4e09\u3001\u7814\u7a76\u65b9\u6cd5 \u672c\u7814\u7a76\u6240\u63d0\u51fa\u4e4b\u65b9\u6cd5\u4e3b\u8981\u53ef\u4ee5\u5206\u70ba\u56db\u5927\u90e8\u5206\uff0c\u4e00\u958b\u59cb\u6703\u63a5\u6536\u6587\u672c\u7684\u8f38\u5165\uff0c\u4e26\u900f\u904e Feature Extraction \u4f86\u5c0d\u8f38\u5165\u7684\u6587\u672c\u9032\u884c Filter Stopwords\u3001Stemming\u3001Lemmatization \u4f86\u53d6\u5f97\u6587\u672c \u7684\u7279\u5fb5\u3002\u4e4b\u5f8c\u5206\u70ba\u5169\u500b\u968e\u6bb5\u9032\u884c\uff0c\u7b2c\u4e00\u968e\u6bb5\u6703\u900f\u904e Topic Calculation\uff0c\u4ee5 LDA \u4e3b\u984c\u6a21\u578b \u4f86\u53d6\u5f97\u5404\u500b\u6587\u672c\u7684\u4e3b\u984c\u7279\u5fb5\uff0c\u4e26\u5c0d\u5404\u500b\u6587\u672c\u505a\u4e3b\u984c\u5206\u4f48\u7684\u6a19\u8a18\uff1b\u7b2c\u4e8c\u968e\u6bb5\u5247\u662f\u900f\u904e WordVec[13]\u5b57\u8a5e\u5411\u91cf\u7a7a\u9593\u6a21\u578b\uff0c\u5c07\u5404\u500b\u6587\u672c\u7684\u5b57\u8a5e\u8f49\u6210\u5411\u91cf\u8868\u793a\u3002\u6700\u5f8c\u5408\u4f75\u7b2c\u4e00\u968e\u6bb5 \u8207\u7b2c\u4e8c\u968e\u6bb5\u7684\u7d50\u679c\uff0c\u900f\u904e\u76e3\u7763\u5f0f\u5b78\u7fd2\u6cd5\u4f86\u8a13\u7df4\u5206\u985e\u5668\u4e26\u53d6\u5f97\u7acb\u5834\u7d50\u679c\u3002 \u5716\u4e00\u3001\u7cfb\u7d71\u67b6\u69cb\u5716 \u9808\u5148\u5c07\u6587\u672c\u7684\u5167\u5bb9\u8f49\u63db\u6210\u5411\u91cf\uff0c\u4f5c\u70ba\u8a13\u7df4\u5206\u985e\u5668\u524d\u7684\u8f38\u5165\u3002\u70ba\u4e86\u8868\u793a\u5b57\u8a5e\u7684\u8a9e\u7fa9\u95dc\u4fc2\uff0c \u672c\u7814\u7a76\u4f7f\u7528\u9810\u5148\u8a13\u7df4\u597d\u7684 Word2Vec \u9032\u884c\u6587\u672c\u7684\u5411\u91cf\u8f49\u63db\uff0c\u4ee5\u4fbf\u8655\u7406\u5f8c\u7e8c\u7684\u5be6\u9a57\u8207\u6b65 \u9a5f\u3002Word2Vec[14]Embedding model\uff0c\u5728 Word2Vec \u4e2d\uff0c\u900f\u904e\u795e\u7d93\u7db2\u8def\u7684\u65b9\u6cd5\uff0c\u5c07\u6587\u672c\u4e2d\u7684\u6587\u5b57\u8207\u6587\u5b57\u7684\u95dc \u4fc2\u8f49\u5316\u6210\u5177\u6709\u8a9e\u7fa9\u95dc\u4fc2\u548c\u8a9e\u6cd5\u7d50\u69cb\u7684\u5411\u91cf\u5f62\u5f0f\u3002Word2Vec \u4e2d\u6709\u5169\u500b\u6a21\u578b\uff0c\u5982\u5716\u4e8c\uff0c\u4e00 \u500b\u70ba\u9023\u7e8c\u578b\u8a5e\u888b\u6a21\u578b(CBOW)\u4ee5\u53ca\u8df3\u8e8d\u5f0f\u6a21\u578b(Skip-gram)\u5169\u7a2e\u3002CBOW \u662f\u900f\u904e\u8f38\u5165\u7684\u4e0a \u4e0b\u6587\u4f86\u9810\u6e2c\u5b57\u8a5e\uff0c\u800c Skip-gram \u662f\u900f\u904e\u8f38\u5165\u7684\u5b57\u8a5e\u4f86\u9810\u6e2c\u4e0a\u4e0b\u6587\u3002 \u5728\u8a13\u7df4\u5b57\u8a5e\u6a21\u578b\u7684\u6642\u9593\u4e0a\uff0cSkip-gram \u76f8\u8f03\u65bc CBOW \u7684\u8a13\u7df4\u6642\u9593\u6703\u8f03\u4e45\uff0c\u4f46\u662f\u5728\u8a9e\u610f\u5206 \u901a\u5e38\u5e36\u6709\u9ad8\u4e3b\u89c0\u6027\u8207\u9ad8\u60c5\u7dd2\u6975\u6027\u7684\u7528\u5b57\uff0c\u800c Atheism \u7684\u8b70\u984c\u5247\u5e38\u51fa\u73fe\u4e00\u4e9b\u60c5\u7dd2\u5b57\u773c\u8f03\u4e0d \u505c\u7528\u8a5e\u904e\u6ffe\u7684\u57fa\u672c\u6587\u672c\u8655\u7406\uff1bLemmatization \u70ba\u8a5e\u578b\u9084\u539f\uff1bStemming \u70ba\u8a5e\u5e79\u63d0\u53d6\u3002 \u5716\u56db\u3001LDA \u6587\u672c\u4e3b\u984c\u5206\u5e03\u6982\u7387\u793a\u610f\u5716 Change \u7565\u70ba\u589e\u52a0\uff1b\u5728 Atheism \u4e0a\u5247\u6c92\u660e\u986f\u589e\u52a0\uff0c\u539f\u56e0\u53ef\u80fd\u5728 Feminist \u8207 Hillary \u7684\u8b70\u984c \u7406\u6587\u672c\uff0c\u4e26\u6bd4\u8f03\u4ed6\u5011\u5c0d\u65bc SVM \u5206\u985e\u7684\u5f71\u97ff\u3002\u4e0b\u5716 4.2 \u5206\u5225\u6709\u4e09\u7a2e\u8655\u7406\u65b9\u6cd5\uff0cNormal \u70ba \u5716\u4e8c\u3001CBOW \u4ee5\u53ca Skip-gram \u4e4b Word Embedding \u6a21\u578b\u793a\u610f\u5716 \u5716\u4e09\u3001LDA \u751f\u6210\u6b65\u9a5f \u4e3b\u984c\u7279\u5fb5\u7684\u64f7\u53d6\u6b65\u9a5f\u70ba\u5716\u56db\u6240\u793a\uff0c\u8f38\u5165\u4e00\u7bc7\u6587\u672c\uff0c\u7d93\u7531 LDA \u53d6\u5f97\u8a72\u6587\u672c\u5728\u5404\u500b\u4e3b\u984c\u4e2d \u7684\u5206\u5e03\u6982\u7387\u3002\u672c\u7814\u7a76\u63a1\u53d6\u6587\u672c\u7684\u4e3b\u984c\u5206\u5e03\u6982\u7387\u4f86\u4f5c\u70ba\u4e3b\u984c\u7279\u5fb5\uff0c\u4e26\u8207\u6240\u7522\u751f\u7684\u5b57\u8a5e\u5411\u91cf \u7279\u5fb5\u9032\u884c\u7d50\u5408\uff0c\u4f86\u9032\u884c\u5f8c\u7e8c\u7684\u5be6\u9a57\u3002 \u5728\u7acb\u5834\u5224\u65b7\u4e0a\uff0c\u63a8\u6587\u7684\u4eba\u5c0d\u65bc\u4e8b\u60c5\u7684\u4e3b\u89c0\u8207\u5ba2\u89c0\u6975\u70ba\u91cd\u8981\uff0c\u5177\u6709\u4e2d\u7acb\u7acb\u5834\u7684\u63a8\u6587\u5177\u6709\u975e \u4e3b\u89c0\u6027\u7684\u770b\u6cd5\uff0c\u800c\u5728\u9019\u908a\u5177\u6709\u652f\u6301\u8207\u53cd\u5c0d\u7684\u8a0a\u606f\u5247\u6703\u6709\u975e\u4e2d\u7acb\u60c5\u7dd2\u7684\u8981\u7d20\uff0c\u5728\u9019\u968e\u6bb5\u5982 \u5716\u4e94\u6240\u793a\uff0c\u56e0\u70ba\u6709\u53ef\u80fd\u6703\u5c0d\u5206\u985e\u6a21\u578b\u7684\u8a13\u7df4\u7522\u751f\u5f71\u97ff\uff0c\u6240\u4ee5\u6211\u5011\u5728\u9019\u5c0d\u6587\u672c\u4f5c\u904e\u6ffe\uff0c\u5c07 \u6587\u672c\u5340\u5206\u6210\u542b\u4e2d\u7acb\u7684\u6587\u672c\u548c\u4e0d\u542b\u4e2d\u7acb\u7684\u6587\u672c\u3002\u672c\u7814\u7a76\u4f7f\u7528 Scikit-learn \u5de5\u5177\u4f86\u9032\u884c SVM \u6975\u611f\u6975\u6027\u5206\u985e\uff0c\u5728\u9032\u884c\u5206\u985e\u524d\uff0c\u5fc5\u9808\u5c07\u8a13\u7df4\u8cc7\u6599\u548c\u6e2c\u8a66\u8cc7\u6599\u8f49\u70ba\u6240\u9700\u7684\u683c\u5f0f\u3002\u5982\u5716\u4e94\uff0c \u6211\u5011\u6240\u8655\u7406\u5b8c\u7684\u6587\u672c\u7d93\u7531\u5411\u91cf\u8f49\u63db\u8207\u4e3b\u984c\u7279\u5fb5\u64f7\u53d6\u7684\u7d50\u5408\uff0c\u900f\u904e SVM \u4f86\u9032\u884c\u60c5\u611f\u6975\u6027 \u7684\u5224\u65b7\u3002 \u56db\u3001\u5be6\u9a57\u65b9\u6cd5 \u672c\u7814\u7a76\u7684\u8cc7\u6599\u96c6\u662f\u4f7f\u7528 SemEval-2016[11]\u5728 Task6 \u4e2d\u6240\u63d0\u4f9b\u7684\u6e2c\u8a66\u8207\u8a13\u7df4\u8cc7\u6599\uff0c\u5171 4063 \u7b46\u3002\u8cc7\u6599\u88e1\u5171\u6709\u4e94\u500b\u4e3b\u984c\u5206\u5225\u70ba\uff1aAtheism\u3001Climate Change\u3001Feminist Movement\u3001Hillary \u5728\u6a5f\u5668\u5b78\u7fd2\u4e0a\uff0c\u6587\u672c\u7684\u8655\u7406\u6975\u70ba\u91cd\u8981\uff0c\u5728\u8f49\u63db\u70ba\u5411\u91cf\u4ee5\u524d\uff0c\u5982\u679c\u4e00\u7bc7\u6587\u672c\u7684\u96dc\u8a0a\u904e\u591a\uff0c \u8207 Hilary Clinton \u5728\u6dfb\u52a0\u4e3b\u984c\u7279\u5fb5\u5f8c\u6709\u660e\u986f\u63d0\u5347\u6e96\u78ba\u7387\uff0c\u5728 Legal. Abortion\u3001Climate \u90a3\u9ebc\u5c31\u6703\u5f71\u97ff\u5230\u5f8c\u7e8c\u7684\u8f38\u5165\u3002\u56e0\u6b64\u5728\u672c\u90e8\u5206\u7684\u5be6\u9a57\uff0c\u6211\u5011\u5c07\u4ee5\u4e0d\u540c\u7684\u6587\u5b57\u8655\u7406\u65b9\u6cd5\u4f86\u8655 \u78ba\u7387\u4e0a\u5927\u591a\u6709\u6240\u63d0\u5347\uff0c\u800c\u4e00\u822c\u7684\u8655\u7406\u8207\u8a5e\u5e79\u63d0\u53d6\u7684\u8cc7\u6599\u591a\u6bd4\u7d93\u7531\u8a5e\u578b\u9084\u539f\u7684\u8cc7\u6599\u7684\u6e96\u78ba \u7387\u4f4e\u3002 \u5f9e\u524d\u9762\u5be6\u9a57\u6211\u5011\u77e5\u9053\u8a5e\u578b\u9084\u539f\u6bd4\u5176\u4ed6\u6587\u672c\u8655\u7406\u7684\u65b9\u6cd5\u597d\uff0c\u56e0\u6b64\u6211\u5011\u5c07\u4f7f\u7528\u8a5e\u578b\u9084\u539f\u7684\u6587 \u672c\u8655\u7406\u65b9\u6cd5\u4f86\u505a\u4e3b\u984c\u7279\u5fb5\u7684\u5be6\u9a57\uff0c\u672c\u5be6\u9a57\u5c07\u628a\u4e3b\u984c\u7279\u5fb5\u8207 Word Embedding \u7684\u5411\u91cf\u7d50 \u5408\uff0c\u9032\u800c\u8207\u672a\u6dfb\u52a0\u4e3b\u984c\u7279\u5fb5\u7684\u4e00\u822c\u5b57\u8a5e\u5411\u91cf\u7279\u5fb5\u505a\u6bd4\u8f03\u3002 \u5716\u4e03\u3001\u4e3b\u984c\u7279\u5fb5\u5be6\u9a57\u6bd4\u8f03\u4e4b\u6e96\u78ba\u7387 \u7531\u5716\u4e03\u5f97\u77e5\uff0c\u6a6b\u8ef8\u70ba\u5404\u500b\u76ee\u6a19\u4e3b\u984c\uff0c\u7e31\u8ef8\u70ba\u6e96\u78ba\u7387\uff0c\u6211\u5011\u53ef\u4ee5\u767c\u73fe Feminist Movement \u7531\u5716\u516b\u5f97\u77e5\uff0c\u6a6b\u8ef8\u70ba\u5404\u500b\u76ee\u6a19\u4e3b\u984c\uff0c\u7e31\u8ef8\u70ba\u6e96\u78ba\u7387\uff0c\u53ef\u4ee5\u77e5\u9053\u4e09\u7a2e\u5206\u985e\u5668\u4e2d\uff0cSVM \u6700\u70ba\u512a\u79c0\uff0c\u800c LSTM \u8207 Bayes \u5206\u985e\u5668\u5247\u8f03\u5dee\u3002\u6839\u64da Igarashi \u7b49\u4eba[11]\u7684\u5be6\u9a57\u8868\u793a\uff0c\u6df1\u5ea6 \u5b78\u7fd2\u7684\u65b9\u6cd5\u4e0d\u898b\u5f97\u6703\u6bd4\u8f03\u597d\u3002 \u4e94\u3001\u7d50\u8ad6 \u672c\u8ad6\u6587\u63d0\u51fa\u4f7f\u7528 Word2Vec \u5b57\u8a5e\u5411\u91cf\uff0c\u7d50\u5408\u4e3b\u984c\u7279\u5fb5\u4f86\u9032\u884c\u7acb\u5834\u6aa2\u6e2c\u3002\u5728\u6587\u672c\u8655\u7406\u65b9\u9762\uff0c \u4f7f\u7528\u4e09\u7a2e\u6587\u672c\u8655\u7406\u65b9\u5f0f\uff0c\u7576\u4e2d\u7684 Lemmatization(\u8a5e\u578b\u9084\u539f)\u65bc\u5be6\u9a57\u4e2d\u5f97\u8b49\uff0c\u5728\u6587\u672c\u5206\u6790\u4e0a \u7684\u6e96\u78ba\u7387\u8f03\u512a\u3002\u5728\u4e3b\u984c\u7279\u5fb5\u7684\u65b9\u9762\uff0c\u900f\u904e\u96b1\u542b\u72c4\u5229\u514b\u96f7\u5206\u5e03(LDA)\u7684\u65b9\u6cd5\u4f86\u7b97\u51fa\u6587\u672c\u7684 \u4e3b\u984c\u5206\u5e03\u80fd\u6709\u6548\u63d0\u5347\u5206\u985e\u6548\u679c\u3002\u5728\u7acb\u5834\u6aa2\u6e2c\u65b9\u9762\uff0c\u672c\u8ad6\u6587\u900f\u904e\u652f\u63f4\u5411\u91cf\u6a5f\u4f86\u8a13\u7df4\u7acb\u5834\u5206 \u985e\u5668\uff0c\u4e26\u4e14\u4f7f\u7528\u5169\u968e\u6bb5\u7684\u65b9\u6cd5\u4f86\u89e3\u6c7a\u4e3b\u89c0\u6027\u7684\u554f\u984c\uff0c\u7d93\u5be6\u9a57\u9a57\u8b49\u6700\u4f73\u7684 F-Measure \u70ba\u4e3b \u6aa2\u6e2c\u7684\u6e96\u78ba\u5ea6\u3002 \u671b\u80fd\u9054\u5230\u66f4\u826f\u597d\u7684\u7d50\u679c\u3002 Clinton\u3001Legal. Abortion \u7b49\uff0c\u5167\u5bb9\u70ba\u63a8\u7279\u4f7f\u7528\u8005\u91dd\u5c0d\u5404\u4e3b\u984c\u8868\u9054\u81ea\u8eab\u7684\u8a55\u8ad6\u6216\u60f3\u6cd5\u3002 \u984c\u76ee\u6a19\u5973\u6b0a\u904b\u52d5\u7684 83.36%\u3002</td></tr><tr><td>\u591a\u6578\u90fd\u662f\u56e0\u70ba\u7acb\u5834\u4e0d\u540c\uff0c\u50cf\u662f\u96fb\u5f71\u8a55\u8ad6\u3001\u7522\u54c1\u610f\u898b\u3001\u7e3d\u7d71\u5927\u9078\u7b49\u6709\u95dc\u7684\u554f\u984c\u3002\u9019\u4e9b\u554f\u984c \u4e00\u500b\u6587\u672c\u662f\u7531\u8a31\u591a\u7684\u6587\u5b57\u6240\u7d44\u6210\u7684\uff0c\u800c\u6587\u672c\u4e2d\u6709\u8a31\u591a\u5b57\u5176\u5be6\u662f\u4e0d\u5fc5\u8981\u7684\uff0c\u5f9e\u6587\u672c\u4e2d\u627e\u5230 \u6790\u4e0a\uff0cSkip-gram \u6bd4 CBOW \u9084\u4f86\u5f97\u597d\uff0c\u672c\u7814\u7a76\u6240\u4f7f\u7528\u7684 Google \u8a13\u7df4\u597d\u7684\u6a21\u578b\uff0c\u662f\u57fa\u65bc \u70ba\u4e86\u627e\u51fa\u6587\u672c\u4f5c\u8005\u5728\u8cbc\u6587\u4e2d\u6240\u96b1\u542b\u7684\u7acb\u5834\uff0c\u6211\u5011\u4f7f\u7528\u6a5f\u5668\u5b78\u7fd2\u7684\u65b9\u5f0f\u4f86\u91dd\u5c0d\u8cbc\u6587\u7684\u5167\u5bb9 \u660e\u986f\u7684\u7528\u5b57\u3002\u70ba\u4e86\u9a57\u8b49\u5206\u985e\u6a21\u578b\u7684\u6548\u679c\uff0c\u6240\u4ee5\u6211\u5011\u4ee5\u539f\u5148\u7684\u8cc7\u6599\u96c6\u9032\u884c\u6e2c\u8a66\uff0c\u4e26\u8207\u6a38\u7d20</td></tr><tr><td>\u901a\u5e38\u90fd\u6703\u88ab\u4eba\u6240\u6536\u96c6\u4f86\u9032\u884c\u63a2\u8a0e\u8207\u5206\u6790\u3002\u65e9\u671f\u7684\u65b9\u6cd5\u901a\u5e38\u662f\u4ee5\u96fb\u8a2a\u6216\u7d19\u672c\u554f\u5377\u4f86\u9032\u884c\u62bd \u6709\u610f\u7fa9\u6216\u662f\u91cd\u8981\u7684\u5b57\uff0c\u5373\u53ef\u7372\u5f97\u8f03\u597d\u7684\u6578\u64da\u4f86\u9032\u884c\u5206\u6790\u3002\u672c\u7814\u7a76\u5f9e\u6587\u672c\u4e2d\u79fb\u9664 Skip-gram \u4e0a\u4f5c\u8a13\u7df4\u7684\u6a21\u578b\u3002\u672c\u7814\u7a76\u4f7f\u7528\u7d93\u904e\u524d\u8655\u7406\u7684\u6587\u672c\uff0c\u4e26\u5229\u7528\u4e3b\u984c\u6a21\u578b\uff0c\u5c0d\u6bcf\u7bc7 \u4f5c\u7acb\u5834\u7684\u5206\u985e\uff0c\u5728\u9019\u908a\u900f\u904e\u5169\u968e\u6bb5\u7684\u65b9\u6cd5\uff0c\u900f\u904e\u627e\u51fa\u8cbc\u6587\u8005\u7684\u4e3b\u89c0\u6027\u7acb\u5834\u8207\u60c5\u611f\u6975\u6027\uff0c \u8c9d\u6c0f\u5206\u985e(Bayes)\u3001\u9577\u77ed\u671f\u8a18\u61b6\u795e\u7d93\u7db2\u8def\u5206\u985e\u6a21\u578b(LSTM)\u9019\u5e7e\u7a2e\u65b9\u6cd5\u4f86\u9032\u884c\u6bd4\u8f03\u3002</td></tr><tr><td>\u6a23\u7684\u8abf\u67e5\uff0c\u6240\u4ee5\u5bb9\u6613\u56e0\u70ba\u4eba\u529b\u8207\u6a23\u672c\u6578\u91cf\u7684\u95dc\u4fc2\uff0c\u9032\u800c\u5f71\u97ff\u5230\u9810\u6e2c\u7684\u7d50\u679c\u3002 Stopwords\uff0c\u63a5\u8457\u4ee5\u8a5e\u5e79\u63d0\u53d6(Stemming)\u3001\u8a5e\u6027\u9084\u539f(Lemmatization)\u7b49\u65b9\u6cd5\uff0c\u4f86\u53d6\u5f97\u6587\u672c \u6587\u672c\u4f5c\u6a19\u8a18\uff0c\u4f86\u53d6\u5f97\u4e3b\u984c\u7279\u5fb5\u3002\u5728\u9019\u4e00\u7bc7\u7ae0\u7bc0\u4e2d\uff0c\u5c07\u6703\u8aaa\u660e\u5982\u4f55\u53d6\u5f97\uff0c\u4e26\u4f7f\u4e4b\u7576\u4f5c\u672c\u7814 \u4f86\u9810\u6e2c\u4f7f\u7528\u8005\u7684\u7acb\u5834\u3002</td></tr><tr><td>\u8fd1\u5e74\u4f86\uff0c\u7db2\u969b\u7db2\u8def\u7684\u666e\u53ca\uff0c\u9020\u5c31\u4e86\u8a31\u591a\u7684\u793e\u7fa4\u7db2\u8def\u5e73\u53f0\u50cf\u662f\uff1aTwitter\u3001Facebook\uff0c\u800c\u6839 \u7684\u5b8c\u6574\u8a9e\u7fa9\u3002\u672c\u7814\u7a76\u4f7f\u7528 NLTK \u4e2d\u7684\u65b7\u8a5e Stopwords \u6e05\u55ae\uff0c\u904e\u6ffe\u6389\u5c6c\u65bc\u505c\u7528\u8a5e\u7684\u6587\u5b57\u4f86 \u7a76\u7279\u5fb5\u3002\u96b1\u542b\u72c4\u5229\u514b\u96f7\u5206\u5e03(Latent Dirichlet Allocation)\u7c21\u7a31 LDA\uff0c\u662f\u7531 Blei[12]\u7b49\u4eba\u5728</td></tr><tr><td>\u64da Statista \u7684\u6578\u64da\u7d71\u8a08[1]\u6307\u51fa\uff1a\u5168\u7403\u793e\u7fa4\u5a92\u9ad4\u7684\u4f7f\u7528\u8005\u5728 2019 \u5e74\u4f30\u8a08\u6709 27.7 \u5104\u4eba\uff0c\u800c \u63d0\u4f9b\u5f8c\u7e8c\u7684\u6b65\u9a5f\u4f5c\u4f7f\u7528\u3002\u8a5e\u5e79\u63d0\u53d6\u662f\u900f\u904e\u62bd\u53d6\u5b57\u8a5e\u7684\u8a5e\u7db4\u4f86\u7372\u5f97\u8a5e\u6839\u7684\u65b9\u5f0f\uff0c\u76ee\u7684\u662f\u8b93 2003 \u5e74\u6240\u63d0\u51fa\u7684\u4e3b\u984c\u8a5e\u888b\u6a21\u578b\u3002\u5229\u7528\u4e0d\u540c\u7684\u6a5f\u7387\u7684\u6f5b\u5728\u4e3b\u984c\uff0c\u4f86\u63cf\u8ff0\u6bcf\u7bc7\u6587\u7ae0\uff0c\u800c\u6bcf</td></tr><tr><td>\u5149\u662f\u5728\u53f0\u7063\uff0c\u793e\u7fa4\u7db2\u7ad9\u7684\u4f7f\u7528\u8005\u6578\u91cf\u5c31\u4f54\u4e86\u7e3d\u4eba\u53e3 89%\uff0c\u80fd\u5f97\u77e5\u793e\u7fa4\u7db2\u7ad9\u5c0d\u65bc\u4eba\u5011\u4f86\u8aaa \u8b8a\u5316\u7684\u5b57\u8a5e\u7c21\u5316\uff0c\u4f7f\u5f97\u6587\u672c\u5206\u6790\u80fd\u7372\u53d6\u5230\u8f03\u597d\u7684\u5b57\u7fa9\u4f86\u4f7f\u7528\uff0c\u672c\u7814\u7a76\u4f7f\u7528\u88ab\u5ee3\u70ba\u63a5\u53d7\u7684 \u4e00\u500b\u4e3b\u984c\u662f\u7531\u5206\u6563\u7684\u4e3b\u984c\u5b57\u8a5e\u6240\u5f62\u6210\u7684\u3002LDA \u4e3b\u984c\u6a21\u578b\u4e5f\u662f\u4e00\u500b\u751f\u6210\u6a21\u578b\uff0c\u4ed6\u5c07\u6bcf\u7bc7</td></tr><tr><td>\u6210\u70ba\u4e86\u751f\u6d3b\u4e2d\u4e0d\u53ef\u6216\u7f3a\u7684\u5b58\u5728\u3002\u4f7f\u7528\u8005\u5e73\u5e38\u6703\u5728\u9019\u4e9b\u5e73\u53f0\u767c\u8868\u81ea\u5df1\u5c0d\u4e0d\u540c\u4e3b\u984c\u7684\u60f3\u6cd5\u6216 Porter Stemmer \u4f86\u9032\u884c\u8a5e\u5e79\u63d0\u53d6\u3002\u8a5e\u578b\u9084\u539f\u80fd\u628a\u4e00\u500b\u4efb\u4f55\u578b\u5225\u7684\u5b57\u8a5e\u9084\u539f\u70ba\u539f\u578b\uff0c\u76ee\u7684 \u6587\u672c\u7684\u4e3b\u984c\u6309\u7167\u4e0d\u540c\u7684\u6a5f\u7387\u4f86\u8868\u73fe\u6bcf\u7bc7\u6587\u7ae0\u3002LDA \u7684\u751f\u6210\u6b65\u9a5f\u5927\u81f4\u4e0a\u70ba\u5716\u4e09\u6240\u793a\uff0c\u8f38 \u5716\u516d\u3001\u4e0d\u540c\u6587\u672c\u8655\u7406\u4e4b\u5206\u985e\u6e96\u78ba\u7387</td></tr><tr><td>\u662f\u610f\u898b\uff0c\u56e0\u6b64\u6bcf\u5929\u90fd\u6703\u6709\u8a31\u591a\u8a0a\u606f\u5b58\u5728\u65bc\u793e\u7fa4\u4e0a\u3002\u82e5\u4ee5\u50b3\u7d71\u7684\u65b9\u6cd5\u4f86\u91dd\u5c0d\u9019\u4e9b\u5927\u91cf\u7684\u8cc7 \u662f\u80fd\u8b93\u5c07\u5b57\u8a5e\u7c21\u5316\u70ba\u6700\u521d\u7684\u5b57\u8a5e\u539f\u5f62\uff0c\u4f7f\u5f97\u6587\u672c\u5206\u6790\u80fd\u7372\u5f97\u5b57\u8a5e\u7684\u5b8c\u6574\u8a9e\u7fa9\uff0c\u7528\u5728\u66f4\u70ba \u5165\u6587\u672c\uff0c\u4e4b\u5f8c\u7d93\u7531 LDA \u8f38\u51fa\u4e3b\u984c\u6578 N \u7684\u5404\u500b\u4e3b\u984c\u3002</td></tr><tr><td>\u8a0a\u4f86\u5c0d\u4f7f\u7528\u8005\u8207\u8cbc\u6587\u9032\u884c\u5206\u6790\uff0c\u9664\u4e86\u9700\u8981\u914d\u7f6e\u8a31\u591a\u4eba\u529b\u8207\u6210\u672c\u82b1\u8cbb\uff0c\u9084\u9700\u8981\u9577\u6642\u9593\u624d\u80fd \u6709\u6240\u7d50\u679c\uff0c\u800c\u4e14\u65b0\u7684\u8a0a\u606f\u9084\u6703\u96a8\u8457\u6642\u9593\u5927\u91cf\u589e\u52a0\uff0c\u56e0\u6b64\u900f\u904e\u6a5f\u5668\u5b78\u7fd2\u7684\u6280\u8853\u8207\u7cfb\u7d71\u81ea\u52d5 \u7cbe\u78ba\u7684\u81ea\u7136\u8a9e\u8a00\u8655\u7406\u4e0a\u7684\u6587\u672c\u5206\u6790\u8207\u8868\u9054\u3002\u5c07\u6587\u672c\u7684\u5167\u5bb9\u900f\u904e\u6a5f\u5668\u5b78\u7fd2\u9032\u884c\u5206\u985e\u524d\uff0c\u5fc5 \u7531\u5716\u516d\u5f97\u77e5\uff0c\u6a6b\u8ef8\u70ba\u5404\u500b\u76ee\u6a19\u4e3b\u984c\uff0c\u7e31\u8ef8\u70ba\u6e96\u78ba\u7387\uff0c\u7d93\u7531\u8a5e\u578b\u9084\u539f\u7684\u8cc7\u6599\uff0c\u5728\u5206\u985e\u7684\u6e96 \u5716\u4e94\u3001Stance Detection \u67b6\u69cb\u5716 \u5716\u516b\u3001\u4e0d\u540c\u5206\u985e\u5668\u4e4b\u5206\u985e\u6e96\u78ba\u7387</td></tr><tr><td>96</td></tr></table>", | |
| "type_str": "table", | |
| "num": null, | |
| "html": null | |
| } | |
| } | |
| } | |
| } |