Varieties of Unranked Tree Languages

We study varieties that contain unranked tree languages over all alphabets. Trees are labeled with symbols from two alphabets, an unranked operator alphabet and an alphabet used for leaves only. Syntactic algebras of unranked tree languages are defined similarly as for ranked tree languages, and an unranked tree language is shown to be recognizable iff its syntactic algebra is regular, i.e., a finite unranked algebra in which the operations are defined by regular languages over its set of elements. We establish a bijective correspondence between varieties of unranked tree languages and varieties of regular algebras. For this, we develop a basic theory of unranked algebras in which algebras over all operator alphabets are considered together. Finally, we show that the natural unranked counterparts of several general varieties of ranked tree languages form varieties in our sense. This work parallels closely the theory of general varieties of ranked tree languages and general varieties of finite algebras, but many nontrivial modifications are required. For example, principal varieties as the basic building blocks of varieties of tree languages have to be replaced by what we call quasi-principal varieties, and we device a general scheme for defining these by certain systems of congruences.