This project was developed as a final group assignment for the Artificial Intelligence course at my bachelor, which I took as a requirement for obtaining my minor degree in Systems Engineering.
Figure 1.- Natural Language Processing illustration. Retrieved from Noun Project and used under a CC-BY 4.0 licence.
Summary
In this project, we captured sentiments of replies to tweets from two Ecuadorian presidential candidates in the 2021 elections (@LassoGuillermo and @ecuarauz). A model of neural networks was used to classify tweets considering their sentiment as positive, negative, or neutral. The training dataset was obtained from the Workshop on Semantic Analysis at SEPLN (TASS) of 2020, 2019, and 2012. Thus, we joined together data of the three editions in one dataset that had information about the id, text, and sentiment associated with tweets. Due to privacy policies, we cannot share the datasets here, but you can register on this page to access all datasets.
The main library used to build our model was Keras and we trained it using Google Cloud GPUs through Google Colab.
Additional information
The complete information regarding model performance metrics, a jupyter notebook and python scripts of reference to reproduce our work, and further details are available on the GitHub repository of this project.