Technical component
mELoNER
Domain-agnostic named-entity recognition and linking for SKG enrichment
đmELoNER (multi-domain entity linking or named-entity recognition) is a set of models designed to identify mentions of key concepts or entities relevant to a research community within the texts of publications and/or other documents available in the corresponding SKG.
These models also normalise the identified mentions to their canonical forms based on specific taxonomies. This ensures consistency and enhances the usability of the extracted information. Each model operates independently and is tailored to specific domain needs. However, the overall framework is built to support seamless integration of entities from new research communities, making it flexible and adaptable for diverse applications.
Functionalities
NER model for core biomedical entities: github.com/sirisacademic/AIObioEnts
Geographical entity recognition and classification of mention intent: github.com/sirisacademic/geordie
Next steps
- NER model for the energy domain: energyType
- NER model for the CCAM domain: vehicleType, sensorType, âŠ
- NER model for the maritime transportation domain: vesselType
- NER model for the neuroscience domain: experimentalApproach, technique, âŠ
- Methods for domain-agnostic NEL given custom taxonomies
