aifeducation (Q5975479)

In social and educational settings, the use of Artificial Intelligence (AI) is a challenging task. Relevant data is often only available in handwritten forms, or the use of data is restricted by privacy policies. This often leads to small data sets. Furthermore, in the educational and social sciences, data is often unbalanced in terms of frequencies. To support educators as well as educational and social researchers in using the potentials of AI for their work, this package provides a unified interface for neural nets in 'keras', 'tensorflow', and 'pytorch' to deal with natural language problems. In addition, the package ships with a shiny app, providing a graphical user interface. This allows the usage of AI for people without skills in writing python/R scripts. The tools integrate existing mathematical and statistical methods for dealing with small data sets via pseudo-labeling (e.g. Lee (2013) <https://www.researchgate.net/publication/280581078_Pseudo-Label_The_Simple_and_Efficient_Semi-Supervised_Learning_Method_for_Deep_Neural_Networks>, Cascante-Bonilla et al. (2020) <doi:10.48550/arXiv.2001.06001>) and imbalanced data via the creation of synthetic cases (e.g. Bunkhumpornpat et al. (2012) <doi:10.1007/s10489-011-0287-y>). Performance evaluation of AI is connected to measures from content analysis which educational and social researchers are generally more familiar with (e.g. Berding & Pargmann (2022) <doi:10.30819/5581>, Gwet (2014) <ISBN:978-0-9708062-8-4>, Krippendorff (2019) <doi:10.4135/9781071878781>). Estimation of energy consumption and CO2 emissions during model training is done with the 'python' library 'codecarbon'. Finally, all objects created with this package allow to share trained AI models with other people.