Research Article

Method of Profanity Detection Using Word Embedding and LSTM

Table 3

Data word embedding [25].

Entire data[‘ieung ah rieul rieul ah-digeut ee nieun ieung yuh ieung hieut wah–ssang-giyok oo rieul jieut ae mieum’, ‘bieup oo tieut ieung ee–giyok oh-nieun ah–nieun ee–hieut uh–hieut ah nieun siot oh giyok ieung ee–ieung oo rieul rieul uh ieung’, ‘chieut oh–ieung ee rieul-giyok giyok ee–bieup ah–rieul oh–siot ee–jieut ah giyok’…]

One-word vector (1D)[0.234, −0.322, 0.401, …, 0.159]

One-sentence vector (2D)[[0, …, 0], [0, …, 0], [0.234, …, 0.159], …, [0.532, …, 0.216]]

Vector of all the sentences (3D)[[[0.234, …, 0.159], [0.532, …, 0.216], …, [0.032, …, 0.659]], [[0.175, …, 0.011], [−0.431, …, 0.382], …, [0.233, …, 0.301]], [[0, …, 0], …, [0.335, …, 0.053], [0.092, …, −0.382]]]