The OPCovidBR is a corpus of Twitter data on COVID-19 annotated with fine-grained opinions and sentiment polarity in Brazilian Portuguese. We extracted 2.000 tweets during the COVID-19 pandemic and annotated them in the fine-grained level opinion, as well as the binary document polarity (positive or negative).
| Polarity Classification | |||||||||
|---|---|---|---|---|---|---|---|---|---|
|
We also provide machine learning-based classifiers for fine-grained opinion and polarity classification tasks using the OPCovidBR dataset. For polarity classification, we tested a cross-domain strategy to measure the performance of the classifiers among different domains. For fine-grained opinion identification, we created a taxonomy of aspects and employed them in conjunction with machine learning methods. Based on the obtained results, we found that the cross-domain method improved the results for the polarity classification task. However, the use of a domain taxonomy presented competitive results for fine-grained opinion identification in Portuguese.
Vargas, F., Dos Santos, R.S.S., Rocha, P.R. (2020). Identifying Fine-Grained Opinion and Classifying Polarity on Coronavirus Pandemic. Proceedings of the 9th Brazilian Conference on Intelligent Systems Intelligent Systems (BRACIS 2020). pp 511–520. Held Online. Brazilian Computing Society (SBC) .
@inproceedings{VargasEtAll2020, author = {Francielle Vargas and Rodolfo Sanches Saraiva Dos Santos and Pedro Regattieri Rocha}, title = {Identifying fine-grained opinion and classifying polarity of twitter data on coronavirus pandemic}, booktitle = {Proceedings of the 9th Brazilian Conference on Intelligent Systems (BRACIS 2020)}, pages = {01-10}, year = {2020}, address = {Rio Grande, RS, Brazil}, crossref = {https://dl.acm.org/doi/abs/10.1007/978-3-030-61377-8_35} }
