The goal of work is to train deep learning models to determine knowledge graph relations (in this case, according to the Freebase schema) that are invoked in user utterances to a conversational system. You will be given a training set of utterances paired with a set of relations, that you can use to train multi-layer perceptron (MLP) models. Here is an example utterance from the dataset:
Show me movies directed by Woody Allen recently. There are two relations that are invoked by this utterance:
movie.directed_by movie.initialreleasedate The task is to train deep learning models and output the associated set of relations when given a new utterance.