A {0, 1}-matrix M is arborescence graphic if there exists an arboresce
nce T such that the arcs of T are indexed on the rows of M and the col
umns of M are the incidence vectors of the are sets of dipaths of T. I
f such a T exists, then T is an arborescence realization for M. This p
aper presents an almost-linear-time algorithm to determine whether a g
iven {0, 1}-matrix is arborescence graphic and, if so, to construct an
arborescence realization. The algorithm is then applied to recognize
a subclass of the extended-Horn satisfiability problems introduced by
Chandru and Hooker (1991).