Word sense disambiguation based on stretchable matching of the semantic template

Wei Wang; Degen Huang; Haitao Yu

doi:10.3934/mfc.2020022

Article Contents

2021, Volume 4, Issue 1: 1-13. Doi: 10.3934/mfc.2020022

This issue Previous Article Next Article On approximation to discrete q-derivatives of functions via q-Bernstein-Schurer operators

Word sense disambiguation based on stretchable matching of the semantic template

1.
School of Computer Science and Technology, Dalian University of Technology, No.2 Linggong Road, Ganjingzi District, Dalian City, Liaoning Province, China
2.
Faculty of Library, Information and Media Science, University of Tsukuba, Tsukuba, Japan

^* Corresponding author: Degen Huang
^* Corresponding author: Degen Huang

Received: January 2020

Revised: August 2020

Early access: September 2020

Published: February 2021

The second author is supported by the National Key Research and Development Program of China (2020AAA0108004) and the National Natural Science Foundation of China under (No. U1936109, 61672127)

Abstract Full Text(HTML) Figure(15) / Table(3) Related Papers Cited by

Abstract

It is evident that the traditional hard matching of a fixed-length template cannot satisfy the nearly indefinite variations in natural language. This issue mainly results from three major problems of the traditional matching mode: 1) in matching with a short template, the context of natural language cannot be effectively captured; 2) in matching with a long template, serious data sparsity will lead to a low success rate of template matching (i.e., low recall); and 3) due to a lack of flexible matching ability, traditional hard matching is more prone to failure. Therefore, this paper proposed a novel method of stretchable matching of the semantic template (SMOST) to deal with the above problems. We have applied this method to word sense disambiguation in the natural language processing field. In the same case of using only the SemCor corpus, the result of our system is very close to the best result of existing systems, which shows the effectiveness of new proposed method.

Keywords:

Mathematics Subject Classification: Primary: 68T50, 68T20; Secondary: 68U15.

Citation:

Full Text(HTML)

Figure 1. One-to-one excellent matching

Download: Full-size image PowerPoint slide

Figure 2. One-to-one poor matching

Download: Full-size image PowerPoint slide

Figure 3. Good matching with stretched template (two random words in test sentence)

Download: Full-size image PowerPoint slide

Figure 4. Good matching with stretched template (two random words in template)

Download: Full-size image PowerPoint slide

Figure 5. Good matching with stretched template (two random words and one obstructing word in test sentence)

Download: Full-size image PowerPoint slide

Figure 6. Words of a test sentence and their sense items

Download: Full-size image PowerPoint slide

Figure 7. A template indexed by the word in a test sentence

Download: Full-size image PowerPoint slide

Figure 8. Matching all word senses in a test sentence for all word senses in the template

Download: Full-size image PowerPoint slide

Figure 9. Ordering of the node numbers of all matched word senses in a test sentence

Download: Full-size image PowerPoint slide

Figure 10. Obtaining the word sense score by the matched node chain

Download: Full-size image PowerPoint slide

Figure 11. Obtaining the final word sense score by the max score

Download: Full-size image PowerPoint slide

Figure 12. Obtaining the template through word sense instead of word

Download: Full-size image PowerPoint slide

Figure 13. Matching a Sense of Word (Algorithm 1)

Download: Full-size image PowerPoint slide

Figure 14. Obtaining the Score of a Matched Node Chain (Algorithm 2)

Download: Full-size image PowerPoint slide

Figure 15. Obtaining the Final Word Sense (Algorithm 3)

Download: Full-size image PowerPoint slide

Table 1. Comparison of F1 scores on our systems with different algorithms on five test sets

Res.	Different algorithms	Sen2	Sen3	Sem07	Sem13	Sem15
SemCor 3.0	SMOST Max.score P1	65.8	63.9	57.6	62.0	65.6
	SMOST Max.score P2	66.3	64.6	57.8	61.7	65.5
	SMOST Max.vote P1	68.0	67.9	59.8	64.2	70.0
	SMOST Max.vote P2	68.8	68.3	60.2	64.2	67.5
	SMOST Max.vote*score P1	67.7	67.1	58.9	64.7	69.2
	SMOST Max.vote*score P2	68.9	68.0	61.1	64.4	66.6

| Show Table

DownLoad: CSV

Table 2. Comparison of F1 scores on several systems using supervised learning method on five test sets

Res.	System	Sen2	Sen3	Sem07	Sem13	Sem15
SemCor 3.0	MFS	65.6	66.0	54.5	63.8	67.1
	IMS baseline(Zhong2010)	70.9	69.3	61.3	65.3	69.5
	BLSTM(Raganato2017)	71.4	68.8	61.8	65.6	69.2
	Seq2Seq(Raganato2017)	68.5	67.9	60.9	64.3	67.3
	SMOST (this paper)	68.9	68.3	61.1	64.7	70.0

| Show Table

DownLoad: CSV

Table 3. Comparison of F1 scores on the systems using template matching method on Sen3 test set

Resource	System	Recall	Precision	F1
multi-res.	SSI (Navigli2004)	68.40	68.50	68.45
	SSI-10words context (Hwang2008)	90.96	57.30	70.31
SemCor2.1	A-RS-10words context(Hwang2008)	56.80	75.53	64.84
+WordNet2.1	SMOST (this paper)	100.0	59.84	74.87

| Show Table

DownLoad: CSV

Related Papers

Cited by

References

[1]	S. W. K. Chan, Generating context templates for word sense disambiguation, AI 2013: Advances in Artificial Intelligence, 8272 (2013), 466-477. doi: 10.1007/978-3-319-03680-9_47.
[2]	D. S. Chaplot and R. Salakhutdinov, Knowledge-based word sense disambiguation using topic models, 32nd AAAI Conference on Artificial Intelligence (AAAI-18), (2018), 1-8.
[3]	X. X. Chen, Z. Y. Liu and M. S. Sun, A United Model for Word Sense Representation and Disambiguation, Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2014 (2014), 1025-1035.
[4]	J. P. Chen and W. Yu, Enriching semantic knowledge for WSD, IEICE Trans, E97-D (2014), 2212-2216. doi: 10.1587/transinf.E97.D.2212.
[5]	Z. Hu, F. Luo and Y. Tan, et al., WSD-GAN: Word sense disambiguation using generative adversarial networks, The Thirty-Third AAAI Conference on Artificial Intelligence (AAAI-19), 33 (2019), 9943-9944. doi: 10.1609/aaai.v33i01.33019943.
[6]	M. Hwang and P. Kim, Adapted relation structure algorithm for word sense disambiguation, Proceedings of Third IEEE International Conference on Digital Information Management (ICDIM), 2008 (2008), 684-688. doi: 10.1109/ICDIM.2008.4746825.
[7]	I. Iacobacci, M. T. Pilehvar and R. Navigli, Embeddings for Word Sense Disambiguation: An Evaluation Study, Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, (2016), 897-907.
[8]	K. L. Jia, Research of word sense disambiguation based on soft pattern, Key Engineering Materials, 460/461 (2011), 130-135. doi: 10.4028/www.scientific.net/KEM.460-461.130.
[9]	M. Kageback and H. Salomonsson, Word sense disambiguation using a bidirectional LSTM, Proceedings of the Workshop on Cognitive Aspects of the Lexicon, (2016), 51-56.
[10]	M. Le, M. Postma, J. Urbani and P. Vossen, A deep dive into word sense disambiguation with LSTM, Proceedings of the 27th International Conference on Computational Linguistics, (2018), 354-365.
[11]	L. Li and Q. Zhou, Chinese word sense disambiguation based on lexical semantic ontology, Journal of Chinese Language and Computing, 18 (2018), 13-23.
[12]	L. L. Li, B. Roth and C. Sporleder, Topic Models for Word Sense Disambiguation and Token-based Idiom Detection, Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics, (2010), 1138-1147.
[13]	W. P Lu, H. Wu, P. Jian, Y. G. Huang and H. Y. Huang, An empirical study of classifier combination based word sense disambiguation, IEICE Trans, E101-D (2018), 225-233. doi: 10.1587/transinf.2017EDP7090.
[14]	F. Luo, T. Liu, Q. Xia, B. Chang and et al., Incorporating glosses into neural word sense disambiguation, Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics, (2018), 2473-2482. doi: 10.18653/v1/P18-1230.
[15]	F. Luo, T. Liu, Z. He and et al., Leveraging gloss knowledge in neural word sense disambiguation by hierarchical Co-attention, Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, (2018), 1402-1411. doi: 10.18653/v1/D18-1170.
[16]	M. Maru, F. Scozzafava, F. Martelli and et al., SyntagNet: Challenging supervised word sense disambiguation with lexical-semantic combinations, Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), (2019), 3532-3538.
[17]	S. Melacci, A. Globo and L. Rigutini, Enhancing modern supervised word sense disambiguation models by semantic lexical resources, 2018 LREC, (2018), 1012-1017.
[18]	O. Melamud, J. Goldberger and I. Dagan, context2vec: Learning generic context embedding with bidirectional lstm, Proceedings of the 20th SIGNLL onference on Computational Natural Language Learning (CoNLL), (2016), 51-61. doi: 10.18653/v1/K16-1006.
[19]	A. Moro, A. Raganato and R. Navigli, Entity linking meets word sense disambiguation: A united approach, Transactions of the Association for Computational Linguistics, 2 (2014), 231-244. doi: 10.1162/tacl_a_00179.
[20]	R. Navigli, Word sense disambiguation: A survey, ACM Computing Surveys, 41 (2009), 1-69. doi: 10.1145/1459352.1459355.
[21]	R. Navigli and P. Velardi, Structural semantic interconnection: A knowledge-based approach to word sense disambiguation, IEEE Transactions on Pattern Analysis and Machine Intelligence, 27 (2005), 1075-1086. doi: 10.1109/TPAMI.2005.149.
[22]	A. R. Pal and D. Saha, Word sense disambiguation: A survey, International Journal of Control Theory and Computer Modeling (IJCTCM), 5 (2015), 1-16.
[23]	A. Panchenko, S. Faralli, S. P. Ponzetto and C. Biemann, Using linked disambiguated distributional networks for word sense disambiguation, Proceedings of the 1st Workshop on Sense, Concept and Entity Representations and their Applications, (2017), 72-78. doi: 10.18653/v1/W17-1909.
[24]	S. Papandrea, A. Raganato and C. D. Bovi, SUPWSD: A flexible toolkit for supervised word sense disambiguation, Proceedings of the 2017 EMNLP System Demonstrations, Association for Computational Linguistics, (2017), 103-108.
[25]	T. Pasini and R. Navigli, Train-O-Matic: Large-Scale Supervised Word Sense Disambiguation in Multiple Languages without Manual Training Data, Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, Association for Computational Linguistics, (2017), 78-88. doi: 10.18653/v1/D17-1008.
[26]	L.N. Pina and R. Johansson, Embedding Senses for Efficient Graph-based Word Sense Disambiguation, Proceedings of the 2016 Workshop on Graph-based Methods for Natural Language Processing, NAACL-HLT 2016, (2016), 1-5.
[27]	A. Popov, Word sense disambiguation with recurrent neural networks, Proceedings of the Student Research Workshop associated with RANLP 2017, (2017), 25-34. doi: 10.26615/issn.1314-9156.2017_004.
[28]	A. Raganato, C. D. Bovi and R. Navigli, Neural sequence learning models for word sense disambiguation, Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, Association for Computational Linguistics, (2017), 1156-1167.
[29]	A. Raganato, J. Camacho-Collados and R. Navigli, Word sense disambiguation: A unified evaluation framework and empirical comparison, Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics, (2017), 99-110.
[30]	B. S. Rintyarna and R. Sarno, Topic models for word sense disambiguation and token-based idiom detection, 2016 Fourth International Conference on Information and Communication Technologies (ICoICT), (2016), 1-5.
[31]	F. Tacoa, D. Bollegala and M. Ishizuka, A context expansion method for supervised word sense disambiguation, 2012 IEEE Sixth International Conference on Semantic Computing, (2012), 339-341. doi: 10.1109/ICSC.2012.27.
[32]	A. Trask, P. Michalak and J. Liu, SENSE2VEC - A fast and accurate method for word sense disambiguation in neural word embeddings, arXiv:1511.06388, Under Review as a Conference Paper at ICLR 2016, (2015), 1{9.
[33]	X. D. Wang, X. R. Tang, W. G. Qu and M. Gu, Word sense disambiguation by semantic inference, 2017 International Conference on Behavioral, Economic, Socio-cultural Computing, (2017), 1-6. doi: 10.1109/BESC.2017.8256391.
[34]	D. Y. Yuan, J. Richardson, R. Doherty, C. Evans and E. Altendorf, Semi-supervised Word Sense Disambiguation with Neural Models, Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers, (2016), 1374-1385.
[35]	C. X. Zhang, L. R. Sun, X. Y. Gao, Z. M. Lu and Y. Yue, Integrate chinese semantic knowledge into word sense disambiguation, International Journal of Hybrid Information Technology, 8 (2015), 105-116.
[36]	C. X. Zhang, L. R. Sun and X. Y. Gao, Determine word sense based on semantic and syntax information, International Journal of Database and Theory and Application, 9 (2016), 17-22.
[37]	Z. Zhong and H. T. Ng, It makes sense: A wide-coverage word sense disambiguation system for free text, Proceedings of the ACL 2010 System Demonstrations, (2010), 78-83.