臺灣博碩士論文加值系統

閱讀新聞可以讓我們獲得各種不同領域的知識，了解社會、政治、經濟、科技和文化等方面的時事和趨勢，也能讓我們在家中就能透過新聞了解世界上各個角落正在發生的事情，培養良好的國際視野並保持與國際社會的連接。科技進步導致新聞平台從紙本轉移到數位媒體，提供即時且多樣化資訊，然而這樣的改變讓我們難以接收所有時事資訊，可能錯過事件背後重要的細節和原因。
為了解決上述問題，我們透過抽取新聞中的事件並分析事件之間的關係，以建構通盤性事件推理知識圖譜，用於下游推理問答任務。最後，我們運用通盤性事件推理知識圖譜實現一個完整原因之推理問答系統，並進一步結合中文寫作中的起承轉合技巧，做為我們的推理結構，讓系統產生的回覆更加完整、簡潔且通順的答案，讓人們在資訊爆炸的時代快速取得所需資訊，滿足求知慾望。

Reading news provides us with knowledge across various domains and helps us understand current issues in society, politics, economy, technology, and culture. It allows us to stay informed about global events, fostering a global perspective and international connections. However, the shift from print to digital media has made it challenging to keep up with all the news, potentially causing us to miss crucial details and underlying reasons behind issues.
To tackle this, we propose extracting issues from news articles and analyzing their relationships to build a comprehensive issue reasoning knowledge graph. This graph can be used for downstream reasoning and question-answering tasks.
Furthermore, we utilize this knowledge graph to develop a long-form question answering system. By incorporating the narrative structure of Chinese writing, including the topic, supporting, transitions, and conclusion (TSRC structure), our system generates comprehensive, concise, and coherent responses. This enables individuals to quickly access the information they seek and satisfy their thirst for knowledge in the era of information overload.

摘要 I
Abstract II
誌謝 III
List of Content IV
List of Tables VI
List of Figures VII
1 Introduction 1
1.1 Background 1
1.2 Motivation 1
1.3 Goal 3
1.4 Method 3
1.5 Contribution 5
2 Related works 6
2.1 Event Extraction 6
2.2 Knowledge graph 7
2.3 Question Answering 8
2.4 Multi-hop Reasoning 10
3 Method 12
3.1 System Framework 12
3.2 Text Preprocessing 14
3.2.1 Data Cleaning 14
3.2.2 Time Normalization 14
3.2.3 Coreference 16
3.3 Reasoning Structure Construction 17
3.3.1 Causal Linking 17
3.3.2 TSRC Structure Construction 18
3.4 Question Analysis 19
3.4.1 Question Concepts Extraction 20
3.5 Reasoning 24
3.5.1 Casual Issue Matching 24
3.5.2 TSRC Structural Reasoning 26
3.6 Answer Generation 28
3.6.1 Reasoning Structure Ranking 28
3.6.2 Long-form Answer Generation 31
4 Experiment 32
4.1 Data 32
4.2 Evaluation metrics 33
4.2.1 Automated Evaluation 33
4.2.2 Human Evaluation 35
4.3 Baseline Models 36
4.4 Single Passage Question Answering on News Data 37
4.4.1 Description 37
4.4.2 Experiment Result 40
4.5 Single Passage Question Answering on DRCD 41
4.5.1 Description 41
4.5.2 Experiment Result 43
4.6 Multiple Passage Question Answering on News Data 43
4.6.1 Description 43
4.6.2 Experiment Result 44
4.7 Error Analysis 44
4.7.1 Hidden Semantic Meaning Error 44
4.7.2 TSRC Structural Reasoning Error 45
4.7.3 Answer Generation Error 47
5 Conclusion 49
6 Reference 50

[Arendarenko and Kakkonen, 2012] Arendarenko, E., & Kakkonen, T. (2012). Ontology-based information and event extraction for business intelligence. In Artificial Intelligence: Methodology, Systems, and Applications: 15th International Conference, AIMSA 2012, Varna, Bulgaria, September 12-15, 2012. Proceedings 15 (pp. 89-102). Springer Berlin Heidelberg.
[Auer et al., 2007] Auer, S., Bizer, C., Kobilarov, G., Lehmann, J., Cyganiak, R., & Ives, Z. (2007, November). Dbpedia: A nucleus for a web of open data. In international semantic web conference (pp. 722-735). Berlin, Heidelberg: Springer Berlin Heidelberg.
[Chang et al., 2020] Chang, W. C., Yu, F. X., Chang, Y. W., Yang, Y., & Kumar, S. (2020). Pre-training tasks for embedding-based large-scale retrieval. arXiv preprint arXiv:2002.03932.
[Chen et al., 2019] Chen, J., Lin, S. T., & Durrett, G. (2019). Multi-hop question answering via reasoning chains. arXiv preprint arXiv:1910.02610.
[Devlin et al., 2018] Devlin, J., Chang, M. W., Lee, K., & Toutanova, K. (2018). Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805.
[Ding et al., 2019] Ding, N., Li, Z., Liu, Z., Zheng, H., & Lin, Z. (2019, November). Event detection with trigger-aware lattice neural network. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) (pp. 347-356).
[Dong and Dong, 2003] Dong, Z., & Dong, Q. (2003, October). HowNet-a hybrid language and knowledge resource. In International conference on natural language processing and knowledge engineering, 2003. Proceedings. 2003 (pp. 820-824). IEEE.
[Ferrada et al., 2017] Ferrada, S., Bustos, B., & Hogan, A. (2017). IMGpedia: a linked dataset with content-based analysis of Wikimedia images. In The Semantic Web–ISWC 2017: 16th International Semantic Web Conference, Vienna, Austria, October 21-25, 2017, Proceedings, Part II 16 (pp. 84-93). Springer International Publishing.
[He and Choi, 2021] He, H., & Choi, J. D. (2021, November). The Stem Cell Hypothesis: Dilemma behind Multi-Task Learning with Transformer Encoders. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing (pp. 5555-5577).
[Hochreiter and Schmidhuber, 1997] Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural computation, 9(8), 1735-1780.
[Jiang et al., 2022] Jiang, Z., Gao, L., Wang, Z., Araki, J., Ding, H., Callan, J., & Neubig, G. (2022, December). Retrieval as Attention: End-to-end Learning of Retrieval and Reading within a Single Transformer. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing (pp. 2336-2349).
[Karpukhin et al., 2020] Karpukhin, V., Oguz, B., Min, S., Lewis, P., Wu, L., Edunov, S., ... & Yih, W. T. (2020, November). Dense Passage Retrieval for Open-Domain Question Answering. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP) (pp. 6769-6781).
[Lin et al., 2004] Lin, C. Y. (2004, July). Rouge: A package for automatic levaluation of summaries. In Text summarization branches out (pp. 74-81).
[Luo et al., 2018] Luo, K., Lin, F., Luo, X., & Zhu, K. (2018). Knowledge base question answering via encoding of complex query graphs. In Proceedings of the 2018 conference on empirical methods in natural language processing (pp. 2185-2194).
[Meyer and Gurevych, 2012] Meyer, C. M., & Gurevych, I. (2012). Wiktionary: A new rival for expert-built lexicons? Exploring the possibilities of collaborative lexicography (pp. 259-291). na.
[Neo4j 2007] Neo4j Graph Database. https://neo4j.com/product/neo4j-graph-database
[OpenAI 2022] Introducing ChatGPT. https://openai.com/blog/chatgpt
[Papineni et al., 2002] Papineni, K., Roukos, S., Ward, T., & Zhu, W. J. (2002, July). Bleu: a method for automatic evaluation of machine translation. In Proceedings of the 40th annual meeting of the Association for Computational Linguistics (pp. 311-318).
[Raffel et al., 2019] Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., ... & Liu, P. J. (2019). Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer. arXiv preprint arXiv:1910.10683.
[Ramponi et al., 2020] Ramponi, A., van der Goot, R., Lombardo, R., & Plank, B. (2020, November). Biomedical event extraction as sequence labeling. In Proceedings of the 2020 conference on empirical methods in natural language processing (emnlp) (pp. 5357-5367).
[Shao et al., 2018] Shao, C. C., Liu, T., Lai, Y., Tseng, Y., & Tsai, S. (2018). DRCD: A Chinese machine reading comprehension dataset. arXiv preprint arXiv:1806.00920.
[Speer et al., 2017] Speer, R., Chin, J., & Havasi, C. (2017, February). Conceptnet 5.5: An open multilingual graph of general knowledge. In Proceedings of the AAAI conference on artificial intelligence (Vol. 31, No. 1).
[Sutton and McCallum, 2012] Sutton, C., & McCallum, A. (2012). An introduction to conditional random fields. Foundations and Trends® in Machine Learning, 4(4), 267-373.
[Valenzuela-Escárcega et al., 2015] Valenzuela-Escárcega, M. A., Hahn-Powell, G., Surdeanu, M., & Hicks, T. (2015, July). A domain-independent rule-based framework for event extraction. In Proceedings of ACL-IJCNLP 2015 System Demonstrations (pp. 127-132).
[Vrandečić and Krötzsch, 2014] Vrandečić, D., & Krötzsch, M. (2014). Wikidata: a free collaborative knowledgebase. Communications of the ACM, 57(10), 78-85.
[Wang et al., 2020] Wang, M., Qi, G., Wang, H., & Zheng, Q. (2020). Richpedia: a comprehensive multi-modal knowledge graph. In Semantic Technology: 9th Joint International Conference, JIST 2019, Hangzhou, China, November 25–27, 2019, Proceedings 9 (pp. 130-145). Springer International Publishing.
[Xu et al., 2021] Xu, W., Zhang, H., Cai, D., & Lam, W. (2021, August). Dynamic Semantic Graph Construction and Reasoning for Explainable Multi-hop Science Question Answering. In Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021 (pp. 1044-1056).
[Yang et al., 2019] Yang, S., Feng, D., Qiao, L., Kan, Z., & Li, D. (2019, July). Exploring pre-trained language models for event extraction and generation. In Proceedings of the 57th annual meeting of the association for computational linguistics (pp. 5284-5294).
[Zhao et al., 2021] Zhao, C., Xiong, C., Boyd-Graber, J., & Daumé III, H. (2021). Multi-step reasoning over unstructured text with beam dense retrieval. arXiv preprint arXiv:2104.05883.