基于整数线性规划重构抽象语义图结构的语义摘要算法

陈鸿昶; 明拓思宇; 刘树新; 高超

doi:10.11999/JEIT180720

基于整数线性规划重构抽象语义图结构的语义摘要算法

doi: 10.11999/JEIT180720

国家数字交换系统工程技术研究中心郑州 450002

基金项目: 国家自然科学基金(61521003)，国家自然科学基金青年科学基金(61601513)

详细信息

作者简介:
陈鸿昶：男，1964年生，教授，博士生导师，研究方向为通信与信息工程、网络大数据

明拓思宇：男，1994年生，硕士生，研究方向为网络大数据、文本摘要

刘树新：男，1987年生，助理研究员，研究方向为网络大数据、复杂网络

高超：男，1982年生，助理研究员，研究方向为网络大数据、计算机视觉

通讯作者:
明拓思宇　1139446336@qq.com

中图分类号: TP391.1
计量
- 文章访问数: 2510
- HTML全文浏览量: 966
- PDF下载量: 103
- 被引次数: 25
出版历程
- 收稿日期: 2018-07-18
- 修回日期: 2018-10-26
- 网络出版日期: 2018-11-19
- 刊出日期: 2019-07-01

Semantic Summarization of Reconstructed Abstract Meaning Representation Graph Structure Based on Integer Linear Pragramming

National Digital Switching System Engineering Technological Research Center, Zhengzhou 450002, China

Funds: The National Natural Science Foundation of China (61521003), The National Natural Science Foundation of China Youth Science Fund (61601513)

摘要

摘要: 针对利用抽象语义(AMR)图来预测摘要子图存在的语义结构不完整问题，该文提出一种基于整数线性规划(ILP)重构AMR图结构的语义摘要算法。首先将数据预处理生成一个AMR总图；然后基于统计特征从AMR总图中抽取出摘要子图重要节点信息；最后利用ILP的方法来对摘要子图中节点关系进行重构，利用完整的摘要子图恢复生成语义摘要。实验结果表明，相比其他语义摘要方法，所提方法的ROUGE值和Smatch值都有显著提高，最多分别提高了9%和14%，该方法有利于提高语义摘要的质量。
- 抽象语义图 /
- 语义摘要 /
- 摘要子图 /
- 语义结构 /
- 整数线性规划
Abstract: In order to solve the incomplete semantic structure problem that occurs in the process of using the Abstract Meaning Representation (AMR) graph to predict the summary subgraph, a semantic summarization algorithm is proposed based on Integer Linear Programming (ILP) reconstructed AMR graph structure. Firstly, the text data are preprocessed to generate an AMR total graph. Then the important node information of the summary subgraph is extracted from the AMR total graph based on the statistical features. Finally, the ILP method is applied to reconstructing the node relationships in the summary subgraph, which is further utilized to generate a semantic summarization. The experimental results show that compared with other semantic summarization methods, the ROUGE index and Smatch index of the proposed algorithm are significantly improved, up to 9% and 14% respectively. This method improves significantly the quality of semantic summarization.
- Abstract Meaning Representation (AMR) graph /
- Semantic summarization /
- Summary subgraph /
- Semantic structure /
- Integer Linear Programming (ILP)

HTML全文

图 1 算法框架图

下载: 全尺寸图片幻灯片

图 2 英文句“I saw Joe’s dog, which was running in the garden”的AMR图表示

下载: 全尺寸图片幻灯片

图 3 AMR图合并生成AMR总图示意图

下载: 全尺寸图片幻灯片

图 4 实验结果AMR图与标准摘要AMR图的对比

下载: 全尺寸图片幻灯片

图 5 L值对摘要质量各指标的影响

下载: 全尺寸图片幻灯片

表 1 摘要子图节点和边预测正确率(%)

	P	R	F1
节点	71.4	82.5	76.5
边	45.6	60.1	51.9

下载: 导出CSV

表 2 不同语义摘要算法的性能对比

算法	ROUGE-1	ROUGE-2	ROUGE-W	Smatch
外部语义资源	20.4	5.6	14.3	17.8
语义聚类	21.2	6.0	15.2	19.1
潜在语义分析	22.8	6.8	14.9	20.5
TextRank算法	25.7	8.1	16.8	24.6
PAS语义图	26.5	9.6	18.6	28.9
本文方法	29.3	10.4	19.6	32.1

下载: 导出CSV

表 3 使用ILP和未使用ILP摘要质量对比

	ROUGE-1	ROUGE-2	ROUGE-W	Smatch
未使用ILP	29.1	9.8	18.7	29.7
使用ILP	29.3	10.4	19.6	32.1
结果提升	0.2	0.6	0.9	2.4

下载: 导出CSV

表 4 与深度学习算法的性能对比

方法	ROUGE-1	ROUGE-2	ROUGE-W	Smatch
本文方法	29.3	10.4	19.6	32.1
深度学习	33.4	13.6	24.8	26.7

下载: 导出CSV

参考文献(16)

LYNN H M, CHOI C, and KIM P. An improved method of automatic text summarization for web contents using lexical chain with semantic-related terms[J]. Soft Computing, 2018, 22(12): 4013–4023. doi: 10.1007/s00500-017-2612-9

SHETTY K and KALLIMANI J S. Automatic extractive text summarization using K-means clustering[C]. International Conference on Electrical, Electronics, Communication, Computer, and Optimization Techniques (ICEECCOT), Mysuru, India, 2017: 1–9.

YU Shanshan, SU Jindian, LI Pengfei, et al. Towards high performance text mining: A TextRank-based method for automatic text summarization[J]. International Journal of Grid and High Performance Computing (IJGHPC) , 2016, 8(2): 58–75. doi: 10.4018/IJGHPC.2016040104

NGUYEN-HOANG T A, NGUYEN K, and TRAN Q V. TSGVi: A graph-based summarization system for Vietnamese documents[J]. Journal of Ambient Intelligence and Humanized Computing, 2012, 3(4): 305–313. doi: 10.1007/s12652-012-0143-x

KHAN A, SALIM N, FARMAN H, et al. Abstractive text summarization based on improved semantic graph approach[J]. International Journal of Parallel Programming, 2018: 1–25. doi: 10.1007/s10766-018-0560-3

BANARESU L, BONIAL C, CAI S, et al. Abstract meaning representation for sembanking[C]. Proceedings of the 7th Linguistic Annotation Workshop and Interoperability with Discourse, Sofia, Bulgaria, 2013: 178–186.

LIU Fei, FLANIGAN J, THOMSON S, et al. Toward abstractive summarization using semantic representations[C]. Proceedings of the 2015 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Denver, USA, 2015: 1077–1086.

SONG Linfeng, PENG Xiaochang, ZHANG Yue, et al. AMR-to-text generation with synchronous node replacement grammar[C]. Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, Vancouver, Canada, 2017: 7–13.

KONSTAS I, IYER S, YATSKAR M, et al. Neural AMR: Sequence-to-sequence models for parsing and generation[C]. Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, Vancouver, Canada, 2017: 146–157.

明拓思宇, 陈鸿昶, 黄瑞阳, 等. 一种基于加权AMR图的语义子图预测摘要算法[J]. 计算机工程, 2018, 44(10): 292–297. doi: 10.19678/j.issn.1000-3428.0050770

MING Tuosiyu, CHEN Hongchang, HUANG Ruiyang, et al. A semantic subgraph predictive summary algorithm based on improved AMR graph[J]. Computer Engineering, 2018, 44(10): 292–297. doi: 10.19678/j.issn.1000-3428.0050770

COLLINS M. Discriminative training methods for hidden markov models: Theory and experiments with perceptron algorithms[C]. Proceedings of the ACL-02 conference on Empirical Methods in Natural Language Processing, Philadelphia, USA, 2002, 10: 1–8.

HERMANN K M, KOČISKÝ T, GREFENSTETTE E, et al. Teaching machines to read and comprehend[C]. Proceeding NIPS’15 Proceedings of the 28th International Conference on Neural Information Processing Systems, Montreal, Canada, 2015, 1: 1693–1701.

LIN Chinyew. ROUGE: A package for automatic evaluation of summaries[C]. Text Summarization Branches Out: Proceedings of the ACL-04 Workshop, Barcelona, Spain, 2004, 10: 74–81.

CAI Shu and KNIGHT K. Smatch: An evaluation metric for semantic feature structures[C]. Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics, Sofia, Bulgaria, 2013, 2: 748–752.

SEE A, LIU P J, and MANNING C D. Get to the point: Summarization with pointer-generator networks[C]. Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, Vancouver, Canada, 2017, 1: 1073–1083.

TAN Jiwei, WAN Xiaojun, and XIAO Jianguo. Abstractive document summarization with a graph-based attentional neural model[C]. Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, Vancouver, Canada, 2017, 1: 1171–1181.

施引文献

期刊类型引用(14)

1.	赵锋，徐志明，刘蕾，艾小锋. 弹道目标特征提取研究现状与展望. 信息对抗技术. 2022(03): 15-32 . 百度学术
2.	徐志明，艾小锋，刘晓斌，吴其华，赵锋. 基于散射中心滑动特性的双基地雷达锥体目标微动特征提取方法. 电子学报. 2021(03): 461-469 . 百度学术
3.	汤华涛，察豪，刘峰. 考虑微多普勒特征的鸟类目标多假设跟踪算法. 舰船电子工程. 2020(05): 43-47 . 百度学术
4.	黄翔东，黎鸣诗，罗蓬，马欣. 基于全相位谱分析的多频内插迭代频率估计器. 振动与冲击. 2019(01): 89-95 . 百度学术
5.	魏嘉琪，张磊，刘宏伟，周叶剑. 基于相位测距的宽带雷达弹道目标微动几何参数估计. 电子与信息学报. 2018(09): 2227-2234 . 本站查看
6.	宋鑫华，叶春茂，杨健. 一种基于雷达窄带回波的转动参数估计方法. 太赫兹科学与电子信息学报. 2018(06): 1087-1092 . 百度学术
7.	李彦兵，张曦文，李飞，陈大庆，高红卫. 一种大加速度机动目标微动参数估计方法. 电子与信息学报. 2017(01): 82-87 . 本站查看
8.	周叶剑，张磊，菅毛，王虹现，邢孟道. 多频调频稀疏分解的微动目标参数估计方法. 电子与信息学报. 2017(10): 2360-2365 . 本站查看
9.	沈菲菲，李立钢，贾配洋，宋振健. 一种基于雷达微多普勒效应苍蝇识别的方法. 电子设计工程. 2017(09): 22-26+30 . 百度学术
10.	曹文杰，赵新燕. 基于三次样条插值的微动目标特征提取. 舰船电子对抗. 2017(04): 55-58 . 百度学术
11.	夏鹏，田西兰. 基于微多普勒信息的弹道目标成像. 雷达科学与技术. 2017(05): 553-557 . 百度学术
12.	肖金国，杜兰，韩勋，曹文杰，刘宏伟. 基于参数化时频分析的进动锥裙目标瞬时微多普勒频率提取方法. 电子与信息学报. 2016(01): 90-96 . 本站查看
13.	黄翔东，王越冬，靳旭康，吕卫. 无窗全相位FFT/FFT相位差频移补偿频率估计器. 电子与信息学报. 2016(05): 1135-1142 . 本站查看
14.	陈春晖，张群，罗迎，孙玉雪. 一种空间微动目标宽带雷达干涉三维成像方法. 电子与信息学报. 2016(12): 3144-3151 . 本站查看