WiRe57 : A Fine-Grained Benchmark for Open Information Extraction
2018-09-24WS 2019Code Available0· sign in to hype
William Léchelle, Fabrizio Gotti, Philippe Langlais
Code Available — Be the first to reproduce this paper.
ReproduceCode
- github.com/rali-udem/WiRe57OfficialIn papernone★ 0
Abstract
We build a reference for the task of Open Information Extraction, on five documents. We tentatively resolve a number of issues that arise, including inference and granularity. We seek to better pinpoint the requirements for the task. We produce our annotation guidelines specifying what is correct to extract and what is not. In turn, we use this reference to score existing Open IE systems. We address the non-trivial problem of evaluating the extractions produced by systems against the reference tuples, and share our evaluation script. Among seven compared extractors, we find the MinIE system to perform best.
Tasks
Benchmark Results
| Dataset | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| WiRe57 | MinIE Gashteovski et al. (2017) | F1 | 35.8 | — | Unverified |
| WiRe57 | ClausIE Del Corro and Gemulla (2013) | F1 | 34.2 | — | Unverified |
| WiRe57 | OpenIE 4 Mausam (2016) | F1 | 26.7 | — | Unverified |
| WiRe57 | Ollie Mausam et al. (2012) | F1 | 23.9 | — | Unverified |
| WiRe57 | ReVerb Fader et al. (2011) | F1 | 20 | — | Unverified |
| WiRe57 | Stanford Angeli et al. (2015) | F1 | 19.8 | — | Unverified |
| WiRe57 | PropS Stanovsky et al. (2016) | F1 | 18.7 | — | Unverified |