For an m-semiring $\mathbb{K}$, a $\mathbb{K}$-relation of arity $k$ is a finite multiset of $k$-tuples each carrying an annotation in $\mathbb{K}$, and a $\mathbb{K}$-instance maps each relation name, at its arity, to a $\mathbb{K}$-relation: the two claims record that the definitions are these ones. An annotated tuple of arity $k$ also reads as a plain tuple of arity $k+1$ over $\mathcal{V} \uplus \mathbb{K}$, its annotation in the last column, which extends to relations and instances, and a composite tuple reads back as an annotated one; this composite reading is what the rewriting of the paper targets. For it, $\mathcal{V} \uplus \mathbb{K}$ is made a value type, with data values below annotations and the annotations compared through the alternative linear order of $\mathbb{K}$.