Task B - Taxonomy Discovery
Discover the taxonomic hierarchy between type pairs
Given instances as
where Ta is the parent (superclass) of Tb, and Tb is the child (subclass) of Ta. The goal is to train a system to correctly identify the taxonomy between type pairs. The training dataset will include term types and taxonomically related type tuples. In the test phase, participants work with just term types and must use their trained models to identify correct taxonomic relationships. The types for the training and test phases will be mutually exclusive. Furthermore, for the testing phase participants will be required to post-process their outputs to return tuples that follow the order of superclass-subclass related types.
SubTask B.1 - Taxonomy Discovery - GeoNames
The GeoNames taxonomy is the classification system used within the GeoNames geographical database. It organizes geographic features and locations into a hierarchical structure, allowing for the systematic categorization of places based on various criteria such as administrative divisions, geographical features, and landmarks. This taxonomy facilitates the retrieval and organization of geographical information, making it easier for users to search, analyze, and utilize data related to locations worldwide. The statistics SubTask B.1 GeoNams is represented as follows:
SubTask B.2 - Taxonomy Discovery - Schema.org
Schema.org is a web content type taxonomy for structured data on the Internet, on web pages, in email messages, and beyond. Schema.org vocabulary can be used to cover entities, relationships between entities, and actions, and can easily be extended through a well-documented extension model. The statistics SubTask B.2 Schema.org is represented as follows:
Schema.org Version: V26.0, 2024-02-12
SubTask B.3 - Taxonomy Discovery - UMLS
The Unified Medical Language System (UMLS) taxonomy is a hierarchical classification within the UMLS knowledge source to organize and categorize medical concepts, terms, and entities. This taxonomy encompasses a wide range of biomedical and health-related information, including diseases, symptoms, drugs, procedures, anatomy, and more. The statistics SubTask B.3 UMLS is represented as follows:
SubTask B.4 - Taxonomy Discovery - GO
Gene Ontology (GO) is a major bioinformatics initiative to unify the representation of gene and gene product attributes across all species. The GO describes knowledge of the biological domain w.r.t three aspects: Biological Process, Molecular Function, and Cellular Component. The statistics SubTask B.4 GO is represented as follows: