Mitigating the Effect of Out-of-Vocabulary Entity Pairs in Matrix Factorization for KB Inference

Abstract

This paper analyzes the varied performance of Matrix Factorization (MF) on the related tasks of relation extraction and knowledge-base completion, which have been unified recently into a single framework of knowledge-base inference (KBI) [Toutanova et al., 2015]. We first propose a new evaluation protocol that makes comparisons between MF and Tensor Factorization (TF) models fair. We find that this results in a steep drop in MF performance. Our analysis attributes this to the high out-of-vocabulary (OOV) rate of entity pairs in test folds of commonly-used datasets. To alleviate this issue, we propose three extensions to MF. Our best model is a TF-augmented MF model. This hybrid model is robust and obtains strong results across various KBI datasets.

Cite

Text

Jain et al. "Mitigating the Effect of Out-of-Vocabulary Entity Pairs in Matrix Factorization for KB Inference." International Joint Conference on Artificial Intelligence, 2018. doi:10.24963/IJCAI.2018/573

Markdown

[Jain et al. "Mitigating the Effect of Out-of-Vocabulary Entity Pairs in Matrix Factorization for KB Inference." International Joint Conference on Artificial Intelligence, 2018.](https://mlanthology.org/ijcai/2018/jain2018ijcai-mitigating/) doi:10.24963/IJCAI.2018/573

BibTeX

@inproceedings{jain2018ijcai-mitigating,
  title     = {{Mitigating the Effect of Out-of-Vocabulary Entity Pairs in Matrix Factorization for KB Inference}},
  author    = {Jain, Prachi and Murty, Shikhar and Mausam,  and Chakrabarti, Soumen},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year      = {2018},
  pages     = {4122-4129},
  doi       = {10.24963/IJCAI.2018/573},
  url       = {https://mlanthology.org/ijcai/2018/jain2018ijcai-mitigating/}
}