This release contains all source code necessary to replicate our
MIML-RE results from the following papers:
Mihai Surdeanu, Julie Tibshirani, Ramesh Nallapati, Christopher D. Manning. Multi-instance Multi-label Learning for Relation Extraction. Proceedings of the 2012 Conference on Empirical Methods in Natural Language Processing and Natural Language Learning (EMNLP-CoNLL), 2012. [pdf] [bib]
Please read the package's README file for details.
We gratefully thank Raphael Hoffmann and Sebastian Riedel for sharing their data and code and for the many helpful discussions. This release includes the data generated by Sebastian Riedel and re-packaged by Raphael Hoffmann (available only under certain conditions - ask mihais AT stanford DOT edu for details). One of our models (edu.stanford.nlp.kbp.slotfilling.classify.HoffmannExtractor) replicates Raphael Hoffmann's best model from his ACL 2011 paper.
We thank the organizers of the TAC-KBP shared tasks for all their effort.
We thank SRI (and in particular Lynn Voss) for being very responsive to Mihai's annoying questions and publicly releasing their gazetteer (faust-gazetteer).