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. The two components are estimated jointly so as to minimize errors in recovering arguments. We study factorization models inspired by previous work in relation factorization and selectional preference modeling. Our models substantially outperform the generative and agglomerative-clustering counterparts and achieve state-of-the-art performance.- Anthology ID:
- Q16-1017
- Volume:
- Transactions of the Association for Computational Linguistics, Volume 4
- Month:
- Year:
- 2016
- Address:
- Cambridge, MA
- Editors:
- Lillian Lee, Mark Johnson, Kristina Toutanova
- Venue:
- TACL
- SIG:
- Publisher:
- MIT Press
- Note:
- Pages:
- 231–244
- Language:
- URL:
- https://aclanthology.org/Q16-1017
- DOI:
- 10.1162/tacl_a_00095
- Cite (ACL):
- Diego Marcheggiani and Ivan Titov. 2016. Discrete-State Variational Autoencoders for Joint Discovery and Factorization of Relations. Transactions of the Association for Computational Linguistics, 4:231–244.
- Cite (Informal):
- Discrete-State Variational Autoencoders for Joint Discovery and Factorization of Relations (Marcheggiani & Titov, TACL 2016)
- PDF:
- https://preview.aclanthology.org/nschneid-patch-4/Q16-1017.pdf
- Code
- diegma/relation-autoencoder
- Data
- New York Times Annotated Corpus