In this paper we present a proposal of a flexible query answering system based on fuzzy logic. We introduce a fuzzy logic programming with a threshold cut and with connectives being arbitrary conjunctors, disjunctors (i.e. they do not need to be associative) and implicators. We use three types of unification: first of syntactical strings based on a similarity relation; second based on linguistic variables; third based on fuzzy conceptual lattices. Our systems assumes a fuzzy logic programming interpreter runs at the remote site. We propose a query dialogue where user can specify thresholds, number of truth values, finite (linear) approximations of fuzzy sets and connectives which are subject of tuning and learning, fuzzy conceptual lattice, fuzzy similarity relations and the form of answer required. This is intended to lower communication (retrieval) complexity. We present a computational (procedural) semantics which is sound and complete wrt. to declarative semantic.