WiRe57 : A Fine-Grained Benchmark for Open Information Extraction
William Léchelle, Fabrizio Gotti, Philippe Langlais (2018), WiRe57 : A Fine-Grained Benchmark for Open Information Extraction, arXiv:1809.08962
Open Information Extraction (OIE) systems, starting with TextRunner, seek to extract all relational tuples expressed in text, without being bound to an anticipated list of predicates. Such systems have been used recently for relation extraction, question-answering, and for building domain-targeted knowledge bases, among others.
Subsequent extractors (ReVerb, Ollie, ClausIE, Stanford Open IE, OpenIE4, MinIE) have sought to improve yield and precision. Despite this, the task definition is underspecified, and, to the best of our knowledge, there is no gold standard.
In order to mitigate this problem, we built 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.
The benchmark data and code is available on Github.