# What exacty is “matrix sparsity” $s$?

In many sources (like on Page 30 here), I found that the complexity of the original Harrow Hassidim Lloyd is stated as $\mathcal{O}(\log (N) s^2 \kappa^2/\epsilon)$ where $s$ is said to be the "matrix sparsity". What is a precise definition of the term "matrix sparsity"? I couldn't find it stated anywhere.