Ecore Metamodels and EcoreBERT Pre-trained Language Model

This dataset contains ecore metamodels from the MAR dataset transformed into tree representations. The original dataset can be found here: http://mar-search.org/experiments/models20/

The data contained in this repository were used to conduct the experiments in the paper: Recommending Metamodel Concepts during Modeling Activities with Pre-Trained Language Models. Link to the paper: https://arxiv.org/abs/2104.01642

The data are organized as follows:

• model : our model trained on the tree representations of metamodels with RoBERTa architecture.
• tokenizers : the byte-level BPE tokenizer we used to train our model.
• train : the training data separated into a training and validation set.
• test : the test data of all experiments conducted in the paper.

This data repository is linked with the following Github repository containing our code: https://github.com/martiwey/metamodel-concepts-bert