Can occlusion facilitate human understanding? Evaluating explainability in named entity recognition

• Alexandre Augusto Aguiar Gomes CEFET-MG
• Leonidas J. F. Braga CEFET-MG
• Marcos P. C. Azevedo CEFET-MG
• Gabriel Assunção UFMG
• Arthur Carvalho Blip
• Michele A. Brandão IFMG
• Daniel H. Dalip CEFET-MG
• Flávio Cardeal Pádua CEFET-MG

DOI: https://doi.org/10.5753/wperformance.2024.2348

Abstract

Explainability techniques help users understand the results of a machine learning model. This work investigates whether the Occlusion explainability technique can generate responses similar to those expected by humans in word classification for Named Entity Recognition. For this, we use a bidirectional LSTM, the CoNLL 2003 dataset, and the manual annotation of 849 sentences, thus creating a reference database. The results show that Occlusion is capable of indicating at least one word that is relevant and compatible with human understanding.

References

Adadi, A. and Berrada, M. (2018). Peeking inside the black-box: A survey on explainable artificial intelligence (xai). IEEE Access, 6:52138–52160.

Arras, L., Montavon, G., Müller, K.-R., and Samek, W. (2017). Explaining recurrent neural network predictions in sentiment analysis. arXiv preprint arXiv:1706.07206.

Brin, S. and Page, L. (1998). The anatomy of a large-scale hypertextual web search engine. Computer networks and ISDN systems, 30(1):107–117.

Danilevsky, M., Qian, K., Aharonov, R., Katsis, Y., Kawas, B., and Sen, P. (2020). A survey of the state of explainable ai for natural language processing. arXiv preprint arXiv:2010.00711.

Goodman, B. and Flaxman, S. (2017). European union regulations on algorithmic decision-making and a “right to explanation”. AI magazine, 38(3):50–57.

Guidotti, R., Monreale, A., Ruggieri, S., Turini, F., Giannotti, F., and Pedreschi, D. (2018). A survey of methods for explaining black box models. ACM computing surveys (CSUR), 51(5):1–42.

Harbecke, D. (2021). Explaining natural language processing classifiers with occlusion and language modeling. arXiv preprint arXiv:2101.11889.

Hu, J. (2018). Explainable deep learning for natural language processing.

Liu, X., Chen, H., and Xia, W. (2022). Overview of named entity recognition. Journal of Contemporary Educational Research, 6(5):65–68.

Pedreshi, D., Ruggieri, S., and Turini, F. (2008). Discrimination-aware data mining. In Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining, pages 560–568.

Pennington, J., Socher, R., and Manning, C. D. (2014). Glove: Global vectors for word representation. In Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP), pages 1532–1543.

Robnik-Šikonja, M. and Kononenko, I. (2008). Explaining classifications for individual instances. IEEE Transactions on Knowledge and Data Engineering, 20(5):589–600.

Russell-Rose, T., Stevenson, M., and Whitehead, M. (2002). The reuters corpus volume 1-from yesterday’s news to tomorrow’s language resources.

Sang, E. F. and De Meulder, F. (2003). Introduction to the conll-2003 shared task: Language-independent named entity recognition. arXiv preprint cs/0306050.

Schweter, S. and Akbik, A. (2020). Flert: Document-level features for named entity recognition. arXiv preprint arXiv:2011.06993.

Siddharthan, A. (2002). Christopher d. manning and hinrich schutze. foundations of statistical natural language processing. mit press, 2000. isbn 0-262-13360-1. 620 pp. 64.95/£44.95(cloth). Natural Language Engineering, 8(1):91–92.

Vajjala, S. and Balasubramaniam, R. (2022). What do we really know about state of the art ner? arXiv preprint arXiv:2205.00034.

Van Lent, M., Fisher, W., and Mancuso, M. (2004). An explainable artificial intelligence system for small-unit tactical behavior. In Proceedings of the national conference on artificial intelligence, pages 900–907. Menlo Park, CA; Cambridge, MA; London; AAAI Press; MIT Press; 1999.

Wallace, E., Feng, S., and Boyd-Graber, J. (2018). Interpreting neural networks with nearest neighbors. In Proceedings of the 2018 EMNLP Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP, pages 136–144, Brussels, Belgium. Association for Computational Linguistics.

Wang, X., Jiang, Y., Bach, N., Wang, T., Huang, Z., Huang, F., and Tu, K. (2020). Automated concatenation of embeddings for structured prediction. arXiv preprint arXiv:2010.05006.

Yamada, I., Asai, A., Shindo, H., Takeda, H., and Matsumoto, Y. (2020). Luke: Deep contextualized entity representations with entity-aware self-attention. arXiv preprint arXiv:2010.01057.

Yang, Z., Dai, Z., Yang, Y., Carbonell, J., Salakhutdinov, R. R., and Le, Q. V. (2019). Xlnet: Generalized autoregressive pretraining for language understanding. Advances in neural information processing systems, 32.