TML is a variant of Datalog. It has a BDD (Binary Decision Diagram) backend and by that, the data is always kept in a compressed form, allowing certain programs to run much faster in comparison to explicit data representation. In some cases, TML allows working over data so big that cannot possibly fit ordinary memory if encoded explicitly.

TML features uncommon optimizations by an implementation of conjunctive query containment algorithms that are not commonly implemented elsewhere. Further, it enhances Datalog in several ways: it allows full first-order logic queries, as well as negation and deletion under the partial fixed-point semantics. It can be used as a generic engine for queries and logical tasks, but its main use case is to be a generic compiler-compiler. To this end, it is coupled with a powerful context-free grammar parser that emits the full parse forest, and then this forest can be modified using TML rules in order to achieve a document in the target language.