It is often the case in considering set differences that all sets considered are subsets of one set.
Let $A$ and $B$ denote sets. In many cases, we take the difference between a set and one contained in it. In other words, we assume that $B \subset A$. In this case, we often take complements relative to the same set $A$. So we do not refer to it, and instead refer to the relative complement of $B$ in $A$ as the complement of $B$.
Let $A$ denote a set, and let $B$ denote a set for which $B \subset A$. We denote the relative complement of $B$ in $A$ by $\relcomplement{B}{A}$. When we need not mention the set $A$, and instead speak of the complement of $B$ without qualification, we denote this complement by $\complement{B}$.
One nice property of a complement when $B \subset A$ is:
Let $E$ denote a set and let $A$ and $B$ denote sets satisfying $A,B \subset E$. Then take all complements with respect to $E$. Here are some immediate consequences of the definition.1