deteccao-ironia

URL: https://github.com/fabio-ricardo/deteccao-ironia

Description: Computer Science Bachelor Thesis.

Research Papers and Thesis

• Submitted Paper: Irony and Sarcasm Detection in Brazilian Portuguese Tweets: A Deep Learning Approach (available soon).
• Bachelor Thesis:
  • Title: Detecção de Ironia e Sarcasmo em Língua Portuguesa: uma abordagem utilizando Deep Learning
  • Publication: Feb/2018
  • DOI: 10.13140/RG.2.2.18896.81924
  • Advisor: Dra. Andreia Gentil Bonfante
  • Author: Fabio Ricardo Araujo da Silva
  • Available here.

Dataset Creation

This repository includes datasets used in both the Bachelor’s thesis and the submitted paper on irony and sarcasm detection in Brazilian Portuguese tweets.

Datasets:

• ironia.csv - 2.52 MB
• nao-ironico.csv - 602 KB

The datasets include labeled tweets for training deep learning models aimed at detecting irony and sarcasm in Brazilian Portuguese. The ironia.csv file contains tweets labeled as ironic, while the nao-ironico.csv file contains non-ironic tweets.

Training Methods

The deep learning approach mentioned in the thesis and paper involves training models on the labeled dataset to detect irony and sarcasm in tweets. Further details on the model architecture and training methodology can be found in the Bachelor’s thesis document.

Results

Specific results are not provided in the repository, but the model has been tested with the mentioned datasets and is available for execution so that users can analyze the results.

The results of the research can be accessed through the Bachelor thesis available here.