This repository contains the code used for the submissions to Task 5. We have participated in both subtasks, as well as in the two available languages: Spanish and English.
Our implementation relies heavily in scikit-learn, pandas, and gsitk.
To install them, please use pip:
pip install scikit-learn pandas gsitk
If you use this code, please consider citing the following papers:
- Oscar Araque, Ignacio Corcuera-Platas, J. Fernando Sánchez-Rada, Carlos A. Iglesias, Enhancing deep learning sentiment analysis with ensemble techniques in social applications, Expert Systems with Applications, Volume 77, 2017, https://doi.org/10.1016/j.eswa.2017.02.002.
@article{ARAQUE2017236,
title = "Enhancing deep learning sentiment analysis with ensemble techniques in social applications",
journal = "Expert Systems with Applications",
volume = "77",
pages = "236 - 246",
year = "2017",
issn = "0957-4174",
doi = "https://doi.org/10.1016/j.eswa.2017.02.002",
url = "http://www.sciencedirect.com/science/article/pii/S0957417417300751",
author = "Oscar Araque and Ignacio Corcuera-Platas and J. Fernando Sánchez-Rada and Carlos A. Iglesias",
keywords = "Ensemble, Deep learning, Sentiment analysis, Machine learning, Natural language processing",
}
- Oscar Araque, Ganggao Zhu, Carlos A. Iglesias, A semantic similarity-based perspective of affect lexicons for sentiment analysis, Knowledge-Based Systems, Volume 165, 2019, https://doi.org/10.1016/j.knosys.2018.12.005.
@article{ARAQUE2019346,
title = "A semantic similarity-based perspective of affect lexicons for sentiment analysis",
journal = "Knowledge-Based Systems",
volume = "165",
pages = "346 - 359",
year = "2019",
issn = "0950-7051",
doi = "https://doi.org/10.1016/j.knosys.2018.12.005",
url = "http://www.sciencedirect.com/science/article/pii/S0950705118305926",
author = "Oscar Araque and Ganggao Zhu and Carlos A. Iglesias",
keywords = "Sentiment analysis, Sentiment lexicon, Semantic similarity, Word embeddings",
}
Unfortunately, the data folder can not be included, since the datasets are from Twitter.