No abstract

No abstract

C FT is a new co nstraint system providing reco rds as logi cal dat.a st ru cture fo r co nstraint (logi c) programming. It can be seen as a genera lizat io n of th e rational tree system employed in Prolog If, where fin er-g rained constr a ints are used , and where subtrees are id entifi ed by key words rath er than by position. C FT is defined by a first-ord er structure consisting of so-called feature trees. Feature trees generali ze th e ordinary trees co rres ponding to first-ord er terms by hav ing th eir edges labeled with fi eld nam es call ed feat ures. The m at hem at ical semanti cs given by the feature trer st ru cture is com plemented with a log ical se m a nt ics given by five axiolll schemes, wh ich we co nj ect ure to co mprise a co mpl ete axiomatization of th e feat ure tree structure. We present a decision m et hod for CFT, wh ich dec id es entailm ent a nd disentailment betwee n possibly existentially quant.ifi ed constraints. Since C FT satisfies th e independ ence property, our dec ision m ethod can a lso be employed for checking the satisfiability of co njunctions of positive a nd negat ive constraints. This includ es quantifi ed negative co nstraints such as VyVz(x 1: f(y , z)). Th e paper also presents an id ealized abstract machin e process ing negat ive a nd positive co nstra ints in crem entally. We argue that an optimized version of the m achin e can decid e satisfiabi li ty and e ntailm ent in quasi-lillear time.

No abstract

No abstract