Abstract
The challenge of climate change and biome conservation is one of the most pressing issues of our time—particularly in Brazil, where key environmental reserves are located. Given the availability of large textual databases on ecological themes, it is natural to resort to question answering (QA) systems to increase social awareness and understanding about these topics. In this work, we introduce multiple QA systems that combine in novel ways the BM25 algorithm, a sparse retrieval technique, with PTT5, a pre-trained state-of-the-art language model. Our QA systems focus on the Portuguese language, thus offering resources not found elsewhere in the literature. As training data, we collected questions from open-domain datasets, as well as content from the Portuguese Wikipedia and news from the press. We thus contribute with innovative architectures and novel applications, attaining an F1-score of 36.2 with our best model.
F. N. Cação and M. M. José—These authors contributed equally to the work.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Articles about the “Environment of Brazil” taken from the following category of Wikipedia: https://pt.wikipedia.org/wiki/Categoria:Meio_ambiente_do_Brasil.
- 2.
Published between January 2018 to June 2021, and scraped from the three biggest newspapers in circulation in Brazil on different topics related to environmental issues in the country (such as deforestation, the status of indigenous peoples).
- 3.
All relevant code, checkpoints, and data (except for the scrapped news, in respect of copyright laws) are available or referenced in the Github repository: https://github.com/C4AI/deepage.
- 4.
The pre-trained model is available at https://huggingface.co/unicamp-dl/ptt5-base-portuguese-vocab.
- 5.
Available at https://dumps.wikimedia.org/.
- 6.
The query we developed is described in the Appendix A.
- 7.
For example, if the word “biome” were a substring of a certain question or its answer, the word “Brazil” or other names of states in the country should also be so, so as not to include QA-pairs about biomes from other countries in our selection.
References
Paschoal, A.F.A., et al.: A Bilingual Portuguese-English dataset for question-answering about the ocean. In: 30th ACM International Conference on Information and Knowledge Management (CIKM 2021) (2021). https://doi.org/10.1145/3459637.3482012
Bahdanau, D., Cho, K.H., Bengio, Y.: Neural machine translation by jointly learning to align and translate. In: 3rd International Conference on Learning Representations, ICLR 2015 - Conference Track Proceedings, pp. 1–15 (2015)
Bhatia, S., Lau, J.H., Baldwin, T.: Automatic Classification of Neutralization Techniques in the Narrative of Climate Change Scepticism, pp. 2167–2175 (2021)
Carmo, D., Piau, M., Campiotti, I., Nogueira, R., Lotufo, R.: PTT5: pretraining and validating the T5 model on Brazilian Portuguese data, pp. 1–12 (2020). http://arxiv.org/abs/2008.09144
Diggelmann, T., Boyd-Graber, J., Bulian, J., Ciaramita, M., Leippold, M.: CLIMATE-FEVER: a dataset for verification of real-world climate claims, pp. 1–16 (2020). http://arxiv.org/abs/2012.00614
Gao, J., Galley, M., Li, L.: Neural approaches to conversational AI. In: ACL 2018–56th Annual Meeting of the Association for Computational Linguistics, Proceedings of the Conference Tutorial Abstracts, pp. 2–7 (2018). https://doi.org/10.18653/v1/p18-5002
Guu, K., Lee, K., Tung, Z., Pasupat, P., Chang, M.W.: REALM: retrieval-augmented language model pre-training. In: 37th International Conference on Machine Learning, ICML 2020 Part F16814, pp. 3887–3896 (2020)
Kim, Y., Sohn, J., Bang, S., Kim, H.: Deep learning-based question answering system for proactive disaster management. In: Proceedings of the 37th International Symposium on Automation and Robotics in Construction (ISARC) (Isarc), 2020 (2020). https://doi.org/10.22260/isarc2020/0181
Kwiatkowski, T., et al.: Natural questions: a benchmark for question answering research. Trans. Assoc. Comput. Linguistics 7, 453–466 (2019). https://doi.org/10.1162/tacl_a_00276
Lewis, P., et al.: Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks. arXiv (NeurIPS) (2020)
Lewis, P., Wu, Y., Liu, L., Minervini, P., Küttler, H., Piktus, A., Stenetorp, P., Riedel, S.: PAQ: 65 Million Probably-Asked Questions and What You Can Do With Them (2021). http://arxiv.org/abs/2102.07033
Luccioni, A., Baylor, E., Duchene, N.: Analyzing Sustainability Reports Using Natural Language Processing (2020). http://arxiv.org/abs/2011.08073
Nguyen, T., Rosenberg, M., Song, X., Gao, J., Tiwary, S., Majumder, R., Deng, L.: MS MARCO: a human generated Machine reading Comprehension dataset. In: CEUR Workshop Proceedings 1773 (Nips 2016), pp. 1–11 (2016)
Raffel, C., et al.: Exploring the limits of transfer learning with a unified text-to-text transformer. arXiv 21, pp. 1–67 (2019)
Robertson, S., Zaragoza, H.: The probabilistic relevance framework: BM25 and beyond, vol. 3 (2009). https://doi.org/10.1561/1500000019
Severo, E.A., De Guimarães, J.C.F., Dellarmelin, M.L.: Impact of the COVID-19 pandemic on environmental awareness, sustainable consumption and social responsibility: evidence from generations in Brazil and Portugal. J. Clean. Prod. 286 (2021). https://doi.org/10.1016/j.jclepro.2020.124947
Varini, F.S., Boyd-Graber, J., Ciaramita, M., Leippold, M.: ClimaText: a dataset for climate change topic detection, pp. 1–13 (2020). http://arxiv.org/abs/2012.00483
Vaswani, A., et al.: Attention is all you need. Advances in Neural Information Processing Systems 2017-Decem(Nips), pp. 5999–6009 (2017)
Wagner Filho, J.A., Wilkens, R., Idiart, M., Villavicencio, A.: The BRWAC corpus: a new open resource for Brazilian Portuguese. In: LREC 2018–11th International Conference on Language Resources and Evaluation, pp. 4339–4344, May 2019
West, T.A.P., Börner, J., Fearnside, P.M.: Climatic benefits from the 2006–2017 avoided deforestation in Amazonian Brazil. Front. Forests Global Change 2(September), 1–11 (2019). https://doi.org/10.3389/ffgc.2019.00052
Acknowledgements
This work was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES, Finance Code 001) and by the Itaú Unibanco S.A., through the Programa de Bolsas Itaú (PBI) of the Centro de Ciência de Dados (C\(^2\)D) of Escola Politécnica of Universidade de São Paulo (USP). We also gratefully acknowledge support from Conselho Nacional de Desenvolvimento Cientiífico e Tecnológico (CNPq) (grants 312180/2018-7 and 310085/2020-9) and the Center for Artificial Intelligence (C4AI-USP), with support by the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP, grant 2019/ 07665-4) and by the IBM Corporation.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendices
7 Appendix A
To filter QA-pairs in PAQ, we created four sets of keywords: M (from “Must have”: “brazil*” and other states names), G (from “Good to have”; e.g. “deforestation”), U (from “Unique expressions”; e.g. “ibama” or “amazon rainforest”) and E (from “to Exclude”; e.g. “soccer”). For a pair to be selected, it must contain, in its question or answer, at least one expression belonging to M or U. If it was from U, it would already be selected; if it was from M, it could not contain any E expressions. Finally, if the QA-pair contains any keyword from G, it should also contain at least one from M, but none from E either.
8 Appendix B
We performed queries associated with the following keywords (here, translated into English): Amazon rainforest, Cerrado, Climate change, Deforestation, Environmental Conduct Adjustment Agreement, Extractive (from “Extractivism”), Forest fires, Funding for conservation, Green economy, Land grabbing, Mining and Protected areas. It should be noted that the decision to exclude some words is only understandable in Portuguese, since there is no reasonable parallel in English, as with the word “fist” for the keyword “Cerrado” - in Portuguese, the expression “punho cerrado” (“clenched fist”) is common, but it is clearly not directly related to the Cerrado biome. Also for more significant results, we eliminated news related to “agrobusiness” in the Green economy keyword and “militias” in the Land grabbing one.
9 Appendix C
To perform the experiments, we leverage a machine with an AMD Ryzen 9 3950X Processor with 32 CPUs, 64 GB of RAM, 2 NVIDIA RTX 3090 GPUs of 24GB each, on an Ubuntu 20 LTS. Under these conditions, the training of each model lasted between 5h and 8h, with the best model being trained in about 7h40. The inference time – that is, the time it took to a model to generate an answer to a single question from the test set – was around 0.006s.
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this paper
Cite this paper
Cação, F.N., José, M.M., Oliveira, A.S., Spindola, S., Costa, A.H.R., Cozman, F.G. (2021). DEEPAGÉ: Answering Questions in Portuguese About the Brazilian Environment. In: Britto, A., Valdivia Delgado, K. (eds) Intelligent Systems. BRACIS 2021. Lecture Notes in Computer Science(), vol 13074. Springer, Cham. https://doi.org/10.1007/978-3-030-91699-2_29
Download citation
DOI: https://doi.org/10.1007/978-3-030-91699-2_29
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-91698-5
Online ISBN: 978-3-030-91699-2
eBook Packages: Computer ScienceComputer Science (R0)