Insights into Analogy Completion from the Biomedical Domain
This work addresses limitations in analogy evaluation for researchers in NLP and biomedicine, but it is incremental as it modifies existing methods rather than introducing a new paradigm.
The paper tackled the problem of evaluating word embeddings through analogy completion by identifying flawed assumptions in standard methodology and proposing modifications to allow multiple correct answers and report additional metrics like MAP and MRR. They introduced BMASS, a biomedical dataset, and showed it poses significant challenges to current methods.
Analogy completion has been a popular task in recent years for evaluating the semantic properties of word embeddings, but the standard methodology makes a number of assumptions about analogies that do not always hold, either in recent benchmark datasets or when expanding into other domains. Through an analysis of analogies in the biomedical domain, we identify three assumptions: that of a Single Answer for any given analogy, that the pairs involved describe the Same Relationship, and that each pair is Informative with respect to the other. We propose modifying the standard methodology to relax these assumptions by allowing for multiple correct answers, reporting MAP and MRR in addition to accuracy, and using multiple example pairs. We further present BMASS, a novel dataset for evaluating linguistic regularities in biomedical embeddings, and demonstrate that the relationships described in the dataset pose significant semantic challenges to current word embedding methods.