Evaluation of Representation Models for Text Classification with AutoML Tools

Automated Machine Learning (AutoML) has gained increasing success on tabular\ndata in recent years. However, processing unstructured data like text is a\nchallenge and not widely supported by open-source AutoML tools. This work\ncompares three manually created text representations and text embeddings\nautomatically created by AutoML tools. Our benchmark includes four popular\nopen-source AutoML tools and eight datasets for text classification purposes.\nThe results show that straightforward text representations perform better than\nAutoML tools with automatically created text embeddings.\n