LiLFeS is a programming system for linguistic formalisms based on typed feature structures. Programming language of LiLFeS is similar to PROLOG and is designed for easy writing of typed feature structures as builtin data structure. LiLFeS can be used as a core system for developing practical natural language processing systems, since it incorporates variables, types, feature value restrictions and equations at the term level. Its syntax also includes disjunction, negation and cut, following the ordinary Prolog syntax. Built-in predicates such as copy of feature structures, assertion, lazy evaluation, multi-dimensional arrays and findall predicates are provided.
LiLFeS can be a core system for practical systems also in terms of its efficiency because LiLFeS provides an engine for efficient execution of programs with feature structure unification. Definite clause programs and feature structure descriptions are compiled into byte code prior to execution. The core engine of the LiLFeS runtime system is an abstract machine, which allows efficient execution of compiled code.