lang2vec
lang2vec copied to clipboard
How is syntactic distance computed when there are missing features
The vectors for syntax have missing entries "--" because the WALS database has missing entries for those features.
Could you please let me know how is syntactic distance computed given these missing entries?
Sorry for the delay in responding, I must have missed the notification email!
The dimensions with missing features are ignored. So if e.g. there are 100 dimensions in each vector, and vector x
is missing features 1-20, and vector y
is missing features 80-100, then the distance is computed over the features 20-80.