The paper experimented with PMI (n-gram-based similarity), FastText (shallow neural model similar to BoW model), LSTM and BERT.
確定! 回上一頁