標籤圖卷積增強式超圖注意力網路之中文健康照護文本多重分類

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：32

、訪客IP：3.145.153.135

姓名

高浩銓(Hao-Chuan Kao) 查詢紙本館藏

畢業系所

電機工程學系

論文名稱

標籤圖卷積增強式超圖注意力網路之中文健康照護文本多重分類
(Label Graph Convolutions Enhanced Hypergraph Attention Networks for Chinese Multi-Label Text Classification in the Healthcare Domain)

相關論文

★ 多重嵌入增強式門控圖序列神經網路之中文健康照護命名實體辨識	★ 基於腦電圖小波分析之中風病人癲癇偵測研究
★ 基於條件式生成對抗網路之資料擴增於思覺失調症自動判別	★ 運用合成器混合注意力改善BERT模型於科學語言編輯
★ 強化領域知識語言模型於中文醫療問題意圖分類	★ 管道式語言轉譯器之中文健康照護開放資訊擷取
★ 運用句嵌入向量重排序器增進中文醫療問答系統效能	★ 利用雙重註釋編碼器於中文健康照護實體連結
★ 聯合詞性與局部語境於中文健康照護實體關係擷取	★ 運用異質圖注意力網路於中文醫療答案擷取式摘要
★ 學習使用者意圖於中文醫療問題生成式摘要	★ 標籤強化超圖注意力網路模型於精神疾病文本多標籤分類

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

至系統瀏覽論文 (2026-9-28以後開放)

摘要(中)

多標籤文本分類目標是自動分析文字內容自動指派一個或多個事先給定的類別標籤，常見的應用包括情感分析、主題檢測及新聞分類等。我們提出一個標籤圖卷積增強式超圖注意力網路 (Label Graph Convolutions Enhanced Hypergraph Attention Networks, LGC-HyperGAT) 模型，藉由超圖注意力網路以找出字詞與句子的關聯，然後用標籤圖卷積網路建構類別標籤之間隱含關係，最後將其銜接在一起，用來預測文本內容種類。實驗資料分為兩個部分，包含 (1) 中文健康照護資料集(HealthDoc)：我們以網路爬蟲蒐集網頁上健康照護相關的新聞、文章專欄以及部落格，並將前處理後的文字內容，由3位大學生人工標記類別標籤，文本總數有2,724篇，平均字數是1,096.91，類別標籤共有9個，分別是疾病資訊、養生保健、心理健康、治療方案、醫療檢測、保健食品、注意事項、藥物以及銀髮族，標籤總數是8,731，平均每篇文章有3.21個標籤。 (2) 中文憂鬱症資料集(PsychPark)：此資料是從心靈園地 (http://www.psychpark.org)網站收集，文本為網友提出的精神疾病狀況與敘述，醫師再依據病患提出的心理問題做多標籤分類，文本總數有2,831篇，平均字數是247.89，類別標籤共有21個，標籤總數是4,425，平均每篇文章有1.56個標籤。藉由實驗結果與錯誤分析得知，我們提出的LGC-HyperGAT模型，在HealthDoc和PsyPark資料集分別達到最好的Macro -F1分數0.725和0.35，比相關研究模型 (CNN, LSTM, Bi-LSTM, FastText, BERT, Graph-CNN, TextGCN, Text-Level-GNN, HyperGAT) 的表現來得更好，藉由錯誤分析可知，標籤分類器學習到的隱含特徵可以有效地提升文本分類的效能。

摘要(英)

Multi-label text classification task focuses on automatically assigning one or more predefined category labels to the text content. The common applications include sentiment analysis, topic detection, news classification, and so on. We propose a Label Graph Convolutions Enhanced Hypergraph Attention Networks (LGC-HyperGAT) model, in which the hypergraph attention networks are used to formulate the relationships between words and sentences in the text content, and the label graph convolutions networks are used to capture the implicit correlations within the labels, and both kinds of networks are finally connected to predict the content labels. There are two experimental datasets including 1) Chinese healthcare dataset (HealthDoc): We firstly crawled to collect health-related news, articles, and blogs on the web. After preprocessing the text content, three undergraduate students were trained to annotate the category manually. A total of 2724 documents were annotated and each contained 1096.91 words on average. There are 9 category labels including disease, health protection, mental health, treatment, examination, ingredient, caution, drug, and elder. The total number of labels is 8,731. Each document contains an average of 3.21 labels. 2) Chinese depression dataset (PsychPark): This data is collected from the PsychPark website (http://www.psychpark.org). Users propose mental illnesses and then doctors classify psychological diseases according to their self-descriptions. The total number of texts is 2,831 and the average number of words is 247.89. The total number of labels is 4,425 across 21 categories with an average of 1.56 labels per document. Based on the experimental results, our proposed LGC-HyperGAT model respectively achieved the best Macro-F1 scores of 0.725 and 0.35 in the HealthDoc and PsyPark datasets, which are better than related models (CNN, LSTM, Bi-LSTM, FastText). , BERT, Graph-CNN, TextGCN, Text-Level-GNN, HyperGAT). Through error analysis, the features learned by the label classifier can effectively improve the performance of multi-label text classification.

關鍵字(中)

★ 嵌入向量
★ 圖神經網路
★ 超圖神經網路
★ 文本分類
★ 健康資訊學

關鍵字(英)

★ embedding
★ graph neural networks
★ hypergraph neural networks
★ text classification
★ health informatics

論文目次

目錄
摘要 i
Abstract ii
致謝 iii
目錄 iv
圖目錄 vi
第一章緒論 1
1-1 研究背景 1
1-2 研究目的 3
1-3 論文架構 4
第二章相關研究 5
2-1 多標籤文本分類 5
2-2 詞嵌入向量 8
2-3 類神經網路 12
2-4 圖神經網路 15
2-5 超圖神經網路 18
第三章模型架構 20
3-1 超圖表示 22
3-2 超邊聚合結構 23
3-3 超圖注意力網路層 26
3-4 標籤相關矩陣 29
3-5 相鄰矩陣特徵傳遞 31
3-6 標籤圖卷積網路層 33
3-7 標籤分類器 34
第四章實驗與結果 35
4-1 實驗資料 35
4-2 實驗設定 45
4-3 詞嵌入向量 47
4-4 評估方法 48
4-5 模型比較 51
4-6 效能分析 63
4-7 錯誤分析 67
第五章結論與未來展望 70
參考文獻 71
附錄 74
文本完整例子 74
文本範例一 74
文本範例二 75
文本範例三 76

參考文獻

參考文獻
[1] Q. Li et al., "A survey on text classification: From shallow to deep learning," arXiv preprint arXiv:2008.00364, 2020.
[2] M. R. Boutell, J. Luo, X. Shen, and C. M. Brown, "Learning multi-label scene classification," Pattern Recognit., vol. 37, no. 9, pp. 1757-1771, 2004.
[3] J. Read, B. Pfahringer, G. Holmes, and E. Frank, "Classifier chains for multi-label classification," Mach. Learn., vol. 85, no. 3, pp. 333-359, 2011.
[4] G. Tsoumakas, I. Katakis, and I. Vlahavas, "Mining multi-label data," in Data mining and knowledge discovery handbook: Springer, 2009, pp. 667-685.
[5] M.-L. Zhang and Z.-H. Zhou, "ML-KNN: A lazy learning approach to multi-label learning," Pattern Recognit., vol. 40, no. 7, pp. 2038-2048, 2007.
[6] A. Elisseeff and J. Weston, "A kernel method for multi-labelled classification," Advances in neural information processing systems, vol. 14, pp. 681-687, 2001.
[7] F. Scarselli, M. Gori, A. C. Tsoi, M. Hagenbuchner, and G. Monfardini, "The graph neural network model," IEEE transactions on neural networks, vol. 20, no. 1, pp. 61-80, 2008.
[8] A. Clare and R. D. King, "Knowledge discovery in multi-label phenotype data," in European conference on principles of data mining and knowledge discovery, 2001: Springer, pp. 42-53.
[9] J. R. Quinlan, C4. 5: programs for machine learning. Elsevier, 2014.
[10] M. J. Berger, "Large scale multi-label text classification with semantic word vectors," Technical report, Stanford University, 2015.
[11] T. Mikolov, K. Chen, G. Corrado, and J. Dean, "Efficient estimation of word representations in vector space," arXiv preprint arXiv:1301.3781, 2013.
[12] J. Pennington, R. Socher, and C. D. Manning, "Glove: Global vectors for word representation," in Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP), 2014, pp. 1532-1543.
[13] A. Krizhevsky, I. Sutskever, and G. E. Hinton, "Imagenet classification with deep convolutional neural networks," Advances in neural information processing systems, vol. 25, pp. 1097-1105, 2012.
[14] Y. Kim, "Convolutional neural networks for sentence classification. EMNLP," ed: Association for Computational Linguistics1746–1751, 2014.
[15] P. Liu, X. Qiu, and X. Huang, "Recurrent neural network for text classification with multi-task learning," arXiv preprint arXiv:1605.05101, 2016.
[16] S. Hochreiter and J. Schmidhuber, "Long short-term memory," Neural computation, vol. 9, no. 8, pp. 1735-1780, 1997.
[17] A. Joulin, E. Grave, P. Bojanowski, and T. Mikolov, "Bag of tricks for efficient text classification," arXiv preprint arXiv:1607.01759, 2016.
[18] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, "Bert: Pre-training of deep bidirectional transformers for language understanding," arXiv preprint arXiv:1810.04805, 2018.
[19] J. Bruna, W. Zaremba, A. Szlam, and Y. LeCun, "Spectral networks and locally connected networks on graphs," arXiv preprint arXiv:1312.6203, 2013.
[20] M. Defferrard, X. Bresson, and P. Vandergheynst, "Convolutional neural networks on graphs with fast localized spectral filtering," Advances in neural information processing systems, vol. 29, pp. 3844-3852, 2016.
[21] L. Yao, C. Mao, and Y. Luo, "Graph convolutional networks for text classification," in Proceedings of the AAAI conference on artificial intelligence, 2019, vol. 33, no. 01, pp. 7370-7377.
[22] L. Huang, D. Ma, S. Li, X. Zhang, and H. Wang, "Text level graph neural network for text classification," arXiv preprint arXiv:1910.02356, 2019.
[23] Y. Feng, H. You, Z. Zhang, R. Ji, and Y. Gao, "Hypergraph neural networks," in Proceedings of the AAAI Conference on Artificial Intelligence, 2019, vol. 33, no. 01, pp. 3558-3565.
[24] S. Bai, F. Zhang, and P. H. Torr, "Hypergraph convolution and hypergraph attention," Pattern Recognit., vol. 110, p. 107637, 2021.
[25] D. M. Blei, A. Y. Ng, and M. I. Jordan, "Latent dirichlet allocation," the Journal of machine Learning research, vol. 3, pp. 993-1022, 2003.
[26] K. Ding, J. Wang, J. Li, D. Li, and H. Liu, "Be more with less: Hypergraph attention networks for inductive text classification," arXiv preprint arXiv:2011.00387, 2020.
[27] Q. Li, Z. Han, and X.-M. Wu, "Deeper insights into graph convolutional networks for semi-supervised learning," in Thirty-Second AAAI conference on artificial intelligence, 2018.
[28] J. Cohen, "A coefficient of agreement for nominal scales," Educational and psychological measurement, vol. 20, no. 1, pp. 37-46, 1960.
[29] J. L. Fleiss, "Measuring nominal scale agreement among many raters," Psychological bulletin, vol. 76, no. 5, p. 378, 1971.
[30] J. R. Landis and G. G. Koch, "The measurement of observer agreement for categorical data," biometrics, pp. 159-174, 1977.
[31] W.-Y. Ma and K.-J. Chen, "Introduction to CKIP Chinese word segmentation system for the first international Chinese word segmentation bakeoff," in Proceedings of the second SIGHAN workshop on Chinese language processing, 2003, pp. 168-171.
[32] T. G. Dietterich, "Approximate statistical tests for comparing supervised classification learning algorithms," Neural computation, vol. 10, no. 7, pp. 1895-1923, 1998.
[33] L. Deng, "The mnist database of handwritten digit images for machine learning research [best of the web]," IEEE Signal Processing Magazine, vol. 29, no. 6, pp. 141-142, 2012.

指導教授

李龍豪(Lung-Hao Lee)

審核日期

2023-3-10

推文