Generating RDF triples from text data
Stepan Zaretckii and Aria Nourbakhsh

In this task we are going to convert natural language sentences to RDFs. For this purpose we consider RDFs triples as a target language and sentences as an input language (here English). We look at this task as the task of machine translation from input language to target. It requires experimenting with different data preprocessing and data augmentation methods over the input sentences. For example, this part requires coming up with a good representation of the sentences and also alignment between RDF triples and corresponding sentences. The resulted input then will be fed to the sequence to sequence neural network. Experimenting with different architectures and parameters of NN is another goal of this research. Finally, we will talk about evaluation metrics that can be defined for this task and compare our results to the baseline.