Approximating Max‐Sum‐Product Problems using Multiplicative Error Bounds