Ontology-style relation annotation datasets: Japanese traffic rules of the road and SemEval-2010 Task 8

Each dataset is in two different formats: (1) Conventional annotation (Normal-RoR stands for the Japanese traffic Rules of the Road) and (2) Ontology-Style Relation (OSR) annotation (OSR-RoR stands for Normal-RoR in OSR format, and OSR-SemEval stands for SemEval-2010 Task 8 in OSR format). --Normal-RoR: Japanese traffic Rules of the Road dataset that are in conventional annotating format using direct links to maintain relationships between entities. ----Normal-RoR_all.json: Contain the whole dataset. ----Folders (1, 2, 3, 4, 5): Contain the five-crossed validating data created from the Normal-RoR.json: ------train.json: Training dataset ------dev.json: Validating dataset ------test.json: Testing dataset ----OSR-RoR: Japanese traffic Rules of the Road dataset that are in Ontology-Style Relation (OSR) annotating format using relation mentions and some RDF standard properties to maintain relationships between entities. ----OSR-RoR_all.json: Contain the whole dataset. ----Folders (1, 2, 3, 4, 5): Contain the five-crossed validating data created from the OSR-RoR.json: ------train.json: Training dataset ------dev.json: Validating dataset ------test.json: Testing dataset ----OSR-SemEval.json: SemEval-2010 Task 8 dataset that are in Ontology-Style Relation (OSR) annotating format using relation mentions and some RDF standard properties to maintain relationships between entities. The differences between the conventional annotation and the OSR are: (1) In the conventional annotation, a direct link is used to maintain relationship between two entities, and (2) In OSR, a relation mention, which is a word or phrase that can represent relationships, is used to maintain relationship between two entities by using “domain” and “range” connecting from the relation mention to the two entity mentions.