Autoencoding for Joint Relation Factorization and Discovery from Text

Diego Marcheggiani, Ivan Titov

Abstract:

We present a method for unsupervised open-domain relation discovery.
In contrast to previous (mostly generative and agglomerative clustering) approaches, our model relies on rich contextual features and
makes minimal independence assumptions.
The model is composed of two parts: a feature-rich relation extractor, which predicts a semantic relation between two entities, and a factorization model, which reconstructs arguments (i.e., the entities) relying on the predicted relation.
We use a variational autoencoding objective and estimate the two components jointly so as to minimize errors in recovering arguments.
We study factorization models inspired by previous work in relation factorization.
Our models substantially outperform the generative and agglomerative-clustering counterparts and achieve state-of-the-art performance.

Conflicts: isti.cnr.it, uva.nl