Fig. 2From: Biomedical semantic indexing by deep neural network with multi-task learningThe CBOW Model and the Skip-gram Model. The CBOW Model uses the previous n words w(t-1)…w(t-n) and the subsequent n words w(t + 1)…w(t + n) to predict the target word w(t). The Skip-gram Model uses the target word w(t) to predict the surrounding words w(t-n)…w(t-1), w(t + 1)…w(t + n)Back to article page