Diachronic Word Embeddings
Diachronic word embeddings are an NLP method for quantitatively tracking semantic change over time. Separate embedding spaces are trained on corpora from distinct historical periods, and alignment techniques such as orthogonal Procrustes transformation are applied to map each period's vectors into a common coordinate system, enabling measurement of lexical semantic drift.
Timeline (last 90 days)
A paper published on arXiv presents a systematic study of the application of diachronic word vector paradigms to Sanskrit.
The basic idea of diachronic word vectors is to train separate vector spaces on corpora from different historical periods, then map them to the same coordinate system via alignment techniques for comparison.
The prerequisite assumptions of the diachronic word vector paradigm—sufficiently large corpora, clear word boundaries, and relatively stable spelling—are often difficult to satisfy in classical languages.
All Facts (3)
A paper published on arXiv presents a systematic study of the application of diachronic word vector paradigms to Sanskrit.
50%UnverifiedThe basic idea of diachronic word vectors is to train separate vector spaces on corpora from different historical periods, then map them to the same coordinate system via alignment techniques for comparison.
50%UnverifiedThe prerequisite assumptions of the diachronic word vector paradigm—sufficiently large corpora, clear word boundaries, and relatively stable spelling—are often difficult to satisfy in classical languages.
50%