First, (i) we detect and extract all the MWTs of the corpus, then (ii) we consider them as single tokens and build embedding vectors based on their contexts as ...
確定! 回上一頁