摘要
针对TempEval-2010会议所提供中文语料中的时序关系识别任务,采用基于条件随机场的方法自动识别获得信号词,并融入跨事件理论,利用基于最大熵模型的分类算法对信号词与其他语言特征进行时序关系识别,同时使用约束传播的推理方法解决语料稀疏问题。实验结果表明,基于条件随机场的方法信号词自动识别准确率为69.21%,融入跨事件理论的时序关系识别准确率达到84.7%,表明所提方法可有效改善识别效果。
For the temporal relation recognition task in Chinese corpus provided by TempEval-2010 conference,this paper uses a method based on conditional random field to automatically identify and obtain the signal words. Fused with cross event theory,it uses classification algorithm of maximum entropy model for temporal relation identification of the words and other linguistic features. Meanwhile,it uses the inference method of constraint propagation to solve the problem of sparse corpus. Experimental results show that the automatic recognition accuracy of signal words by using the method based on conditional random field is 69. 21%, and the recognition accuracy of the temporal relation by method fused with cross event theory is 4. 7%,which means that the proposed method can improve the recognition effect.
引文
[1]Allen J F.Maintaining Knowledge About Temporal Intervals[J].Communications of the ACM,1983,26(11):832-843.
[2]Allen J F.Towards a General Theory of Action and Time[J].Artificial Intelligence,1984,23(2):123-154.
[3]Mani I,Verhagen M,Wellner B,et al.Machine Learning of Temporal Relations[C]//Proceedings of the 21st International Conference on Computational Linguistics and the 44th Annual Meeting of the Association for Computational Linguistics.Stroundsburg,USA:Associa-tion for Computational Linguistics,2006:753-760.
[4]Chambers N.Navytime:Event and Time Ordering from Raw Text[C]//Proceedings of the 7th International Workshop on Semantic Evaluation.Atlanta,Georgia:[s.n.],2013:73-77.
[5]Chambers N,Cassidy T,Mc Dowell B,et al.Dense Event Ordering with a Multi-Pass Architecture[J].Transactions of the Association for Computational Linguistics,2014,2:273-284.
[6]Laokulrat N,Miwa M,Tsuruoka Y,et al.UTTime:Temporal Relation Classification Using Deep Syntactic Features[J].Transactions of the Association for Computational Linguistics,2013,1:88-92.
[7]Yoshikawa K,Riedel S,Asahara M,et al.Jointly Identifying Temporal Relations with Markov Logic[C]//Proceedings of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP.Singapore:[s.n.],2009:405-413.
[8]Kerr C,Hoare T,Marecek J,et al.Integer Programming Ensemble of Classifiers for Temporal Relations[EB/OL].[2016-03-01].https://arxiv.org/pdf/1412.1866v2.pdf.
[9]Li Wenjie,Wong K.A Word-based Approach for Modeling and Discovering Temporal Relations Embedded in Chinese Sentences[J].ACM Transactions on Asian Language Information Processing,2002,1(3):173-178.
[10]Li Wenjie,Wong K,Cao Guihong.Applying Machine Learning to Chinese Temporal Relation Resolution[C]//Proceedings of the 42nd Annual Meeting of the Association for Computational Linguistics.[S.l.]:Association for Computational Linguistics,2004:582-588.
[11]Cheng Yuchang,Asahara M,Matsumoto Y.Constructing a Temporal Relation Tagged Corpus of Chinese Based on Dependency Structure[C]//Proceedings of the 21st Annual Conference of the Japanese Society for Artificial Intelligence.M iyazaki,Japan:[s.n.],2007.
[12]Cheng Yuchang,Asahara M,Matsumoto Y.Use of Event Types for Temporal Relation Identification in Chinese Text[C]//Proceedings of the 6th SIGHAN Workshop on Chinese Language Processing.Hyderabad,India:[s.n.],2008:31-38.
[13]Gupta P,Ji Heng.Predicting Unknown Time Argument Based on Cross-event Propagation[C]//Proceedings of ACL-IJCNLP’09.Singapore:[s.n.],2009:369-372.
[14]Liao Shasha,Grishman R.Using Document Level Crossevent Inference to Improve Event Extraction[C]//Proceedings of the 48th Annual M eeting of the Association for Computational Linguistics.Stroundsburg,USA:Association for Computational Linguistics,2010:789-797.
[15]侯立斌,李培峰,朱巧明,等.基于跨事件理论的缺失事件角色填充研究[J].计算机科学,2012,39(7):200-204.
[16]王凤娥,谭红叶,钱揖丽,等.基于最大熵的句内时间关系识别[J].计算机工程,2012,38(4):37-39.