WN18RR

WN18RR is derived from WN18, with data removed to eliminate test-set leakage due to inverse relations. WN18RR contains 93003 triples, connecting 40943 entities via 11 relations.<br><br>Info about WN18: This WORDNET TENSOR DATA consists of a collection of triplets (synset, relation_type, triplet) extracted from WordNet 3.0 (http://wordnet.princeton.edu). This data set can be seen as a 3-mode tensor depicting ternary relationships between synsets.<br>All *.txt files contain one triplet per line, with 2 synset_ids and relation type identifier in a tab separated format. The first element is the synset_id of the left hand side of the relation triple, the third one is the synset_id of the right hand side and the second element is the name of the type of relations between them.<br>The WN18RR.zip file contains the other files, with more compression than the default "download all".<br>

WN18RR 衍生自 WN18，为消除因反向关系引发的测试集泄露问题，移除了部分数据。WN18RR 共包含 93003 个三元组（triple），通过 11 种关系连接 40943 个实体。

关于 WN18 的说明：本词网张量数据集（WORDNET TENSOR DATA）由从 WordNet 3.0（http://wordnet.princeton.edu）中提取的、形如（同义词集（synset）、关系类型（relation_type）、三元组（triple））的集合构成。该数据集可被视为刻画同义词集之间三元关系的三阶张量。
所有 *.txt 文件每行存储一个三元组，以制表符分隔的格式包含 2 个同义词集 ID 与关系类型标识符。其中首个元素为关系三元组左侧的同义词集 ID，第三个元素为右侧的同义词集 ID，第二个元素为二者间的关系类型名称。
WN18RR.zip 压缩包包含其余文件，其压缩率高于默认的"全部下载"模式。