(SC) MINIMUM SET COVER (hypergraph formulation)
Instance: Hyper graph G=(V,E), for each vertex v in V a weight W(v)>=0;
Solution: A set cover for G, i.e., a subset C of V such that,
          for each edge (v,u) in E, at least one of u and v are in C;
Measure : W(C) (i.e. Sum {W(v): v in C})

Define k=Max{|e| : e in E}

+----------------------------------------------------------------+
|   The "Local-Ratio Theorem" (oriented for SC)                  |
+----------------------------------------------------------------+
|   If   C is k-approximation SC w.r.t the weight function W1    |
|   and  C is k-approximation SC w.r.t the weight function W2    |
|   then C is k-approximation SC w.r.t the weight function W1+W2 |
+----------------------------------------------------------------+

Let {v1,v2,...vk} be any hyper-edge s.t. epsilon>0 where:
   epsilon = Min(W(v1),W(v2),...,W(vk))
define W1(v1)=W1(v2)=...=W1(vk)=epsilon and 0 for all others.

Q1: All set covers are k-approximations w.r.t W1. Why?

The algorithm: iteratively subtract W1 as above.
Induction hyp. implies that reclusive call with W2=W-W1
would return a set cover C which is d-approximation w.r.t. W2.
(Where d is an upper bound on the edge cardinality)
C (and any set cover) is d-approximation w.r.t. W1 (see Q1).
By the "Local-Ratio Theorem" C is d-approximation w.r.t. W=W1+W2.

Q2: What to do with zero weight vertices?

Hence: The following linear time algorithm is d-approximation.

     +-------------------------------------------------+
     |  Algorithm Bar-Yehuda&Even(G,W)                 |
     +-------------------------------------------------+
     |     for each edge e in E                        |
     |        Let epsilon = Min{W(v): v endpoint of E} |
     |        W(v) -= epsilon for all v endpoint of E  |
     |     return {v: W(v)=0}                          |
     +-------------------------------------------------+