In a seminal paper, Charikar et al. derive upper and lower bounds on the approximation ratios for several grammar-based compressors, but in all cases there is a gap between the lower and upper bound. Here the gaps for LZ78 and BISECTION are closed by showing that the approximation ratio of LZ78 is Θ((n/log n)2/3), whereas the approximation ratio of BISECTION is Θ(√(n/log n)). In addition, the lower bound for RePair is improved from Ω(√(log n)) to Ω(log n/log log n). Finally, results of Arpe and Reischuk relating grammar-based compression for arbitrary alphabets and binary alphabets are improved.

Many XML documents are data-centric and do not make use of the inherent document order. Can we provide stronger compression for such documents through giving up order? We first consider compression via minimal dags (directed acyclic graphs) and study the worst case ratio of the size of the ordered dag divided by the size of the unordered dag, where the worst case is taken for all trees of size n. We prove that this worst case ratio is n / log n for the edge size and n log log n / log n for the node size. In experiments we compare several known compressors on the original document tree versus on a canonical version obtained by length-lexicographical sorting of subtrees. For some documents this difference is surprisingly large: reverse binary dags can be smaller by a factor of 3.7 and other compressors can be smaller by factors of up to 190.

A grammar-compressed ranked tree is represented with a linear space overhead so that a single traversal step, i.e., the move to the parent or the ith child, can be carried out in constant time. The data structure is extended so that equality of subtrees can be checked in constant time.