Services on Demand
Journal
Article
Indicators
- Cited by SciELO
- Access statistics
Related links
- Similars in SciELO
Share
RISTI - Revista Ibérica de Sistemas e Tecnologias de Informação
Print version ISSN 1646-9895
Abstract
ACOSTA, Rubén Sánchez; VILLEGAS, Claudio Meneses and NORAMBUENA, Brian Keith. Heuristics for Data Augmentation in NLP: Application to scientific paper reviews. RISTI [online]. 2019, n.34, pp.44-53. ISSN 1646-9895. https://doi.org/10.17013/risti.34.44-53.
Data augmentation techniques are essential for training machine learning algorithms, where the initial data set is smaller than required due to the model complexity. In machine learning models, the robustness of the training process is highly dependent on large volumes of labeled data, which are expensive to produce. An effective approach to deal with this problem is to automatically generate new tagged examples using data augmentation techniques. In the processing of natural language, particularly in the Spanish language, there is a lack of well-defined techniques that allow increasing a set of data. In this article, we propose a set of heuristics for data augmentation in NLP, which are applied to the domain of reviews of scientific articles.
Keywords : Data Augmentation; NLP; Paper Reviews.