结合Hybrid Attention机制和BiLSTM-CRF的汉语否定语义表示及标注

doi:10.3778/j.issn.1002-8331.2201-0088

摘要/Abstract

摘要： 阅读理解中否定是一种复杂的语言现象，其往往会反转情感或态度的极性。因此，正确分析否定语义对语篇理解具有重要意义。现有否定语义分析方法存在两个问题：第一，研究的否定词较少达不到应用目的；第二，目前汉语否定语义标注只是标注整个句子，这无法明确否定语义。针对该问题提出基于汉语框架语义知识库（Chinese FrameNet）进行否定语义角色标注方法。在框架语义学理论指导下结合汉语否定语义特征对已由FrameNet继承的否定框架重新构建；为了解决捕捉长距离信息以及句法特征问题，提出一种基于Hybrid Attention机制的BiLSTM-CRF语义角色标注模型，其中，Hybrid Attention机制层将局部注意与全局注意结合准确表示句子中的否定语义，BiLSTM网络层自动学习并提取语句上下文信息，CRF层预测最优否定语义角色标签。经过比对验证，该模型能够有效提取出含有否定语义信息，在否定语义框架数据集上F1值达到89.82%。

关键词: 汉语框架语义知识库, 语义角色标注, 否定框架, 双向长短期记忆网络, 混合注意力机制

Abstract: Negation is a complex language phenomenon in reading comprehension, which often reverses the polarity of emotion or attitude. Therefore, the correct analysis of negative semantics is of great significance to discourse understanding. There are two problems in the existing negative semantic analysis methods：first, there are few negative words, which can not achieve the purpose of application. Second, at present, Chinese negative semantic tagging is only tagging the whole sentence, which can not clarify the negative semantics. To solve this problem, a negative semantic role annotation method based on Chinese FrameNet is proposed. Firstly, under the guidance of frame semantics theory, combined with the semantic characteristics of Chinese negation, the negative framework inherited by FrameNet is reconstructed. Secondly, in order to solve the problem of capturing long-distance information and syntactic features, a BiLSTM-CRF semantic role annotation model based on Hybrid Attention mechanism is proposed. The Hybrid Attention mechanism layer combines local attention and global attention to accurately represent the negative semantics in the sentence, the BiLSTM network layer automatically learns and extracts the sentence context information, and the CRF layer predicts the optimal negative semantic role label. Through comparison and verification, the model can effectively extract negative semantic information, and the F1 value reaches 89.82% on the negative semantic framework data set.

Key words: Chinese framework knowledge base, semantic role annotation, negation frame, bi-directional long short-term memory（BiLSTM）, hybrid attention mechanism

李晋荣, 吕国英, 李茹, 柴清华, 王超. 结合Hybrid Attention机制和BiLSTM-CRF的汉语否定语义表示及标注[J]. 计算机工程与应用, 2023, 59(9): 167-175.

LI Jinrong, LYU Guoying, LI Ru, CHAI Qinghua, WANG Chao. Chinese Negative Semantic Representation and Annotation Combined with Hybrid Attention Mechanism and BiLSTM-CRF[J]. Computer Engineering and Applications, 2023, 59(9): 167-175.

参考文献

[1] OGDEN C K，RICHARDS I A.The meaning of meaning：a study of the influence of language upon thought and of the science of symbolism[M].New York：Harvest/HBJ，1989.
[2] 王霜梅.语义学发展轨迹及其滞后原因探究——从语言学的意义转向谈起[J].学习与探索，2009（3）：190-192.
WANG S M.Research on the development track of semantics and the reasons for its lag—from the meaning turn of linguistics[J].Study and Exploration，2009（3）：190-192.
[3] 管春林.汉英否定对比研究[M].天津：中国书籍出版社，2020.
GUAN C L.A contrastive study of negation in Chinese and english[M].Tianjin：China Books Publishing House，2020.
[4] 龙波，郭文.基于FrameNet的汉语语义框架网络自动构造[J].现代计算机，2010（2）：4-7.
LONG B，GUO W.Automatic construction of Chinese FrameNet based on frameNet[J].Modern Computer，2010（2）：4-7.
[5] YOU L，LIU K.Building Chinese framenet data-base[C]//IEEE International Conference on Natural Language Processing and Knowledge Engineering，2005：301-306.
[6] 李茹.汉语句子框架语义结构分析技术研究[D].太原：山西大学，2012.
LI R.Research on frame semantic structure analysis technology for Chinese sentences[D].Taiyuan：Shanxi University，2012.
[7] 袁毓林.汉语否定表达的认知研究和逻辑分析[M].北京：商务印书馆，2018.
YUAN Y L.A cognitive study and logic analysis of the negative expressions in mandarin Chinese[M].Beijing：Commercial Press，2018.
[8] 袁毓林.论否定句的焦点、预设和辖域歧义[J].中国语文，2000（2）：99-108.
YUAN Y L.On the focus，presupposition and scope ambiguity of negative sentences[J].Chinese Language，2000（2）：99-108.
[9] HUANG Z，XU W，YU K.Bidirectional LSTM-CRF models for sequence tagging[J].Computer Science，2015.
[10] XU M Z，YANG B S，ZHANG Y，et al.Leveraging local and global patterns for self-attention networks[C]//Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics，Florence，Italy，2019：3069-3075.
[11] 张谊生.现代汉语副词探索[M].上海：学林出版社，2004.
ZHANG Y S.Exploration of modern chinese adverbs[M].Shanghai：Xuelin Publish House，2004.
[12] 车录彬.现代汉语羡余否定构式研究[D].成都：四川大学，2011.
CHE L B.A research on redundant negation phenomenon in modern Chinese[D].Chengdu：Sichuan University，2011.
[13] 范晓，文贞惠.现代汉语否定范畴研究[D].上海：复旦大学，2003.
FAN X，WEN Z H.A study on negation in modern Chinese[D].Shanghai：Fudan University，2003.
[14] 袁毓林.动词内隐性否定的语义层次和溢出条件[J].中国语文，2012（2）：99-104.
YUAN Y L.The semantic level and overflow condition of implicit negation in verbs[J].Chinese Language，2012（2）：99-104.
[15] 尹洪波.否定与转折[J].语言研究集刊，2014（2）：25-35.
YIN H B.Negation and turning[J].Journal of Linguistics，2014（2）：25-35.
[16] GILDEA D，JURAFSKY D.Automatic labeling of semantic roles[J].Computational Linguistics，2002，28（3）：245-288.
[17] ALESSANDRO M，DANIELE P，ROBERTO B.Tree ker-nels for semantic role labeling[J].Computational Lin-guistics，2008，34（2）：193-224.
[18] YANG B S，MITCHELL T M.A joint sequential and relational model for frame semantic parsing[C]//Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing，2017：1247-1256.
[19] 李济洪.汉语框架语义角色的自动标注技术研究[D].太原：山西大学，2010.
LI J H.Research on techniques of automatic semantic role labeling of Chinese FrameNet[D].Taiyuan：Shanxi University，2010.
[20] 屠寒非，李茹.基于主动学习的汉语框架语义角色标注[D].太原：山西大学，2016.
TU H F，LI R.Active learning for Chinese frame semantic roles labeling[D].Taiyuan：Shanxi University，2016.
[21] 党帅兵.基于词分布表征的汉语框架语义角色识别研究[D].太原：山西大学，2015.
DANG S B.Research on identifying of CFN semantic roles based on distributed word representation[D].Taiyuan：Shanxi University，2015.
[22] 王瑞波，李济洪，李国臣.基于Dropout正则化的汉语框架语义角色识别[J].中文信息学报，2017，31（1）：147-154.
WANG R B，LI J H，LI G C，et al.Chinese FrameNet semantic role identification based on dropout regular-ization[J].Journal of Chinese Information Processing，2017，31（1）：147-154.
[23] QIAN F，SHA L，CHANG B B，et al.Syntax aware LSTM model for semantic role labeling[C]//Proceedings of the 2nd Workshop on Structured Prediction for Natural Language Processing，2017.
[24] 王晓辉，李茹，王志强，等.基于Self-Attention的句法感知汉语框架语义角色标注[C]//中国计算语言学大会，2020：616-623.
WANG X H，LI R，WANG Z Q，et al.Syntax-aware Chinese frame semantic role labeling based on self-attention[C]//Proceedings of the 19th Chinese National Conference on Computational Linguistics，Haikou，China，2020：616-623.
[25] 黄恩任，齐沪扬.现代汉语否定结构的认知研究[D].上海：上海师范大学，2015.
HUANG E R，QI H Y.Cognitive research on negative structure in modern Chinese[D].Shanghai：Shanghai Normal University，2015.
[26] 蔡永强.“连…都/也…”结构的认知考察[D].北京：北京语言文化大学，2002.
CAI Y Q.A cognitive investigation of the structure of “Even…All/Also…”[D].Beijing：Beijing Language and Culture University，2002.