Documents usually contain mentions of people, locations, and organizations. The goal is to identify these mentions and map them to correct entities in a knowledge base (such as Wikipedia).
CoNEREL, a named entity recognition and linking system.
Related publications:
Biomedical concepts often appear in medical documents under different name variations (synonyms). We aim to calculate vector representations for these biomedical names such that names of the same concept will have similar representations, hence, are close to each other in the embedding space.
PCA visualization of 254 name embeddings. These embeddings belong to 10 disease concepts (shown in different colors)
Our approach
Baseline: Average word embedding
Related publications: