Existing natural language processing systems are vulnerable to noisy inputs
resulting from misspellings. On the contrary, humans can easily infer the
corresponding correct words from their misspellings and surrounding context.
Inspired by this, we address the stand-alone spelling correction