https://github.com/NPCai/Open-IE-Papers

Open-IE-Papers

Table of Contents

  1. General
  2. Literature Reviews
  3. Papers - Neural Networks
  4. Papers - Parse-based and statistical
  5. Papers - Older papers and legacy systems
  6. Training and Testing Data

General

This README containts OpenIE and ORE papers and resources. Summaries are by @jbecke and @TheodoreChristakis, to the best of our abilities after reading each paper or testing the system (when available). We welcome pull requests with additional resources, papers, or data.

Literature Reviews

[A Survey on Open Information Extraction](https://arxiv.org/abs/1806.05599). Most up-to-date literature review (June 2018), convering non-neural network based approaches to OpenIE. Whereas I've classified by age in this document, the authors classify by method of extraction (learning-based, rule-based, clause-based, inter-propositional).

[Creating a Large Benchmark for Open Information Extraction](http://www.aclweb.org/anthology/D16-1252): summarizes the field and creates a benchmark for OpenIE systems and creates the first large benchmark dataset (note: not large enough to train NN's).

[Effectiveness and Efficiency of Open Relation Extraction](http://www.anthology.aclweb.org/D/D13/D13-1043.pdf): Review of the limited work done in the field of ORE (open relation extraction).

Papers - Neural Networks

[Neural Open Information Extraction](https://arxiv.org/pdf/1805.04270.pdf): AFAIK, the first use of ANNs (seq2seq with attention) applied to OpenIE. Author bootstrapped tuples from high-confidence OpenIE-4 and makes the data available. However, the data isn't very clean; a quick glance shows a lot of malformed/incorrect tuples.

[Supervised Open Information Extraction](http://aclweb.org/anthology/N18-1081): expands on the idea of turning QA datasets into OpenIE datasets. Trains an ANN with using an interesting feature representation, uses seq2seq model to generate BIO tags and then creates tuples from that using a deterministic algorithm.