Table 1. Korean dependency parsing algorithm.

Algorithm: Korean dependency parsing
Input: a sequence of words in a sentenceOutput: a list of governors and dependency relations
stack, buffer ← make_stack(), make_buffer(eojeols)action ← make_list()buffer ← reverse(buffer)
pred_head ← make_list(size=len(eojeols) + 1)pred_rel ← make_list(size=len(eojeols) + 1)
for i in range(len(eojeols) * 2)
   if is_empty(buffer) then
      end algorithm
   next_action, relation ← oracle(stack, buffer, action)
   if next_action = 'RIGHT-ARC' then
      action.push(next_action)
   else if next_action = 'REDUCE' then
      action.push(next_action)
      pred_head[buffer.top()] ← stack.top()
      pred_rel[buffer.top()] ← relation
      stack.push(buffer.pop())
   else
      raise parse_error
end forreturn pred_head, pred_rel