What is “rest known data” for imputation?

I'm familiar with most methods of imputation, such as mean imputation, and co-prediction (same as imputation using regression).

But, in this article, on page 2 it mentions "rest known data to impute missing terms".

I have never heard of this method and I'm wondering what it means. I know how to use the $X^2$ test to test for independence, so that analysis is sound.....But, how does one use that for actually filling in the missing values? and what is "rest known data to impute missing terms"?

• I imagine it is a typo for rest of the known. – mdewey Dec 14 '16 at 9:22
• But how does one use the "rest [of the] known data" to impute the missing ones? – guimption Dec 17 '16 at 16:19