Hassan At-Kaci. LOGIN: A Logic Programming Language With Built-In Inheritance. Tech. Rep., Microelectronics and Computer Technology Corporation, 1985.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....attributes. Feature logic has its origin in the three areas of knowledge representation with concept descriptions, frames, or y terms [13, 34, 35, 1] natural language processing, in particular approaches based on unification grammars [26, 24, 45, 43, 39, 42] and constraint (logic) programming [3, 5, 28, 47]. An interesting recent application lies in software configuration management, where feature logic is used to denote software versions and to deduce their mutual consistency [50, 51] The first mathematical treatment of record descriptions was the formalisms of y terms [1] In other approaches, ....

Hassan At-Kaci and Roger Nasr. LOGIN: A logic programming language with built-in inheritance. The Journal of Logic Programming, 3(3):185--215, 1986.

.... useful for constraint simplification [41] during record type inference, but it is also prerequisite for a possible usage of FT in modern constraint programming languages with advanced control mechanisms such as delaying, coroutining, synchronization, committed choice and local computation spaces [2, 5, 47, 38]. We furthermore show that FT has the independence property if the set of features provided by the signature is infinite. Thanks to the independence property, the entailment test is sufficient for testing conjunctions j :j 1 : j n for satisfiability (namely, by testing that none of the ....

Hassan At-Kaci and R. Nasr. LOGIN: A logic programming language with built-in inheritance. Journal on Lisp and Symbolic Computation, 2:51--89, 1989.

.... as record like data structures in constraint (logic) programming [4] 28] and as models of feature descriptions in computational linguistics [7] 6] The use of record like structures in logic programming languages, in the form of socalled terms [1] was pioneered by the languages LOGIN [2] and LIFE [3] More recently, Oz [17, 26] uses a feature constraint system, the semantics of which is directly based on feature trees. In computational linguistics, feature structures have a long history in the field of unification grammars (as described in [25] On leave to: L.R.I. Bat. 490, ....

Hassan Ait-Kaci and Roger Nasr. LOGIN: A logic programming language with built-in inheritance. Journal of Logic Programming, 3:185--215, 1986.

....complexity, second order monadic logic. 1 Introduction Feature logic is a formalism to describe objects by the values of their attributes or features. It has its roots in the three areas of knowledge representation, with concept descriptions, frames, or y terms [Brachman Levesque, 1984, At Kaci, 1986, Nebel, 1990, Nebel Smolka, 1990] natural language processing, especially approaches based on unification grammars [Kay, 1979, Kaplan Bresnan, 1982, Shieber et al. 1983, Shieber, 1986, Pollard Sag, 1994, Rounds, 1997] and constraint programming languages with record structures [At Kaci ....

.... 1986, Nebel, 1990, Nebel Smolka, 1990] natural language processing, especially approaches based on unification grammars [Kay, 1979, Kaplan Bresnan, 1982, Shieber et al. 1983, Shieber, 1986, Pollard Sag, 1994, Rounds, 1997] and constraint programming languages with record structures [At Kaci Nasr, 1986, Mukai, 1988, At Kaci Podelski, 1993, Smolka, 1995] Two main approaches to feature logics can be distinguished according to the underlying logical structures. Both approaches rely on quite similar syntactic constructions called feature constraints but differ significantly in their semantics. ....

Hassan At-Kaci & Roger Nasr. LOGIN: A logic programming language with built-in inheritance. The Journal of Logic Programming, 3(3):185--215, 1986.

....other words, its purpose is not only for compilers to analyze programs but also for programmers to understand their programs better. Efficiency. Mode analysis is a constraint satisfaction problem with many simple mode constraints, and can be solved efficiently by unification over feature graphs [1]. Modularity. Thanks to its incremental nature, it is naturally amenable to separate analysis of large programs. Generality. It allows simple and general formulations of various interesting applications including the diagnosis of non well moded programs [5] and automated debugging [2] We have ....

....possibly shared and f 1 ( means that f( is never shared. The annotations are managed as follows: 1. Annotations in program clauses and initial goal clauses are given according to how the structures are implemented. For instance, suppose two processes p and q in a goal clause : p([1,2,3],X) q( 1,2,3] Y) share a single list [1,2,3] in an ac 3 ffl fi (BF ) p) shared, for a function symbol f occurring at p in B, LV ) For a linear variable with two occurrences at p 1 and p 2 , 8q 2 PTerm (m(p 1 q) in (p 1 q) shared ) p 2 q) shared) when p 1 is a head ....

[Article contains additional citation context not shown here]

Ait-Kaci, H. and Nasr, R., LOGIN: A Logic Programming Language with Built-In Inheritance. J. Logic Programming, Vol. 3, No. 3 (1986), pp. 185--215.

....be shared by other paths; such nodes do not become garbage and must be (re )labeled properly. On the time complexity, we first note that a practical algorithm for the unification of two feature graphs achieves almost linear time complexity with respect to the size l of the graphs to be unified [1][8] That is, the time complexity is O(l 1ff(l) where ff is the inverse of the Ackermann function. The unification algorithm represents the sharing of a node using invisible pointers (which are implicit in Figures 1 3) managed by the union find algorithm. Access to a shared node may involve the ....

.... n log n to the total cost is expected to be much smaller than that of nd1ff(n) 18 For the class of programs being considered, the total correctness of the above analysis algorithm is almost immediate from the total correctness of the underlying unification algorithm for feature graphs [1]. 2.5.3 General Cases How can we cope with clauses with guard goals and or variables with three or more channel occurrences When clauses may have guard goals, Constraint (GV) must be considered as well. We have not fully studied the general and efficient treatment of guard goals, but in ....

Ait-Kaci, H. and Nasr, R., LOGIN: A Logic Programming Language with Built-In Inheritance. J. Logic Programming, Vol. 3, No. 3 (1986), pp. 185--215.

....knowledge structuring. Truly, one can use stream communication to implement delegation, a dynamic form of knowledge sharing (see for example [11, 19, 30] but many useful forms of knowledge structuring fall without this schema, like, for instance, subtyping with complex types as characterized in [2]. The main innovation of LO with respect to Prolog will be in allowing program formulae with multiple heads , i.e. having the form (Head 1 : Headn ) Body: where each Head i is a positive literal. We shall see how the connective acts as a form of disjunction. LO is from this point ....

.... Among the efforts which have been focusing on the dynamic aspects of object oriented programming (i.e. change of state) we have mentioned [11, 17, 19, 29, 30, 31, 32] efforts which have been focusing on the static aspects (i.e. inheritance) of object oriented programming are described in [2, 9, 12, 16, 21, 33]. The framework presented here should provide a theoretical basis for the integration of these two traditions. 10, 28, 8] provide also extensions of Prolog in the direction of object oriented and distributed programming where formulae with multiple heads are allowed; some of the ideas ....

H. Ait-Kaci and R. Nasr. LOGIN: a logic programming language with built-in inheritance. Journal of Logic Programming, 3(3):187--215, 1986.

.... Truly, one can use stream communication to implement a form of inheritance (known as delegation) among classes of objects (see for example [6, 14, 23] but many useful forms of knowledge structuring fall without this schema, like, for instance, subtyping with complex types as characterized in [1]. The main innovation of LO with respect to Prolog will be in allowing program clauses with multiple heads , i.e. having the form (Head 1 : Headn ) Body: where each Head i is a positive literal. We shall see how the connective acts as a form of disjunction. LO is from this point of ....

.... Among the efforts which have been focusing on the dynamic aspects of object oriented programming (i.e. change of state) we have mentioned [6, 12, 14, 22, 23, 24, 25] efforts which have been focusing on the static aspects (i.e. inheritance) of object oriented programming are described in [1, 4, 7, 11, 26]. The framework presented here should provide a theoretical basis for the integration of these two traditions. 5] provides also an extension of Prolog in the direction of object oriented programming where clauses with multiple heads are allowed; some of the ideas characterizing this approach are ....

H. Ait-Kaci and R. Nasr. LOGIN: a logic programming language with built-in inheritance. Journal of Logic Programming, 3(3):187--215, 1986.

....used to organize classes in hierarchies. Object identifiers (oid s) are associated withobjects, in order to allow duplicates and for object sharingand inheritance. The first proposals in this field go backtothe early eighties, and are concerned with the languages LOGIN (Ait Kaci and Nasr [4]) and O logic (Maier [25] LOGIN is a Prolog based language for querying complex object databases. Objects are structures builtbymeans of ordinary logic programmingfunction symbols, and isahierarchies andinheritance are allowed among classes of objects. Thesemantics of the language is developed ....

H. Ait-Kaci and R. Nasr. LOGIN a logic programminglanguage with built-in inheritance. Journal of Logic Programming, 3:185--215, 1986.

....in (the program transformation process that will produce) the new compiled form of the rule. R0: intersectsort X[0] vp testfeature X[0] ARGS, X[1] 1, W1, vp intersectsort X[1] cons S1 testinstlist X[1] 2, W2 getfirst X[1] X[2] intersectsort X[2] verb testfeature X[2] HEAD, X[3], 3, W3, verb R1: testfeature X[2] OBJECT, X[4] 3, W4, verb intersectsort X[4] np R2: testfeature X[2] SUBJECT, X[5] 3, W5, verb R3: getrest X[1] X[6] S2 intersectsort X[6] cons testinstlist X[6] 3, W7 unifyfirst X[6] X[4] testNILrest X[6] S3 R4: unifyfeature X[0] HEAD, ....

.... 3, W3, verb R1: testfeature X[2] OBJECT, X[4] 3, W4, verb intersectsort X[4] np R2: testfeature X[2] SUBJECT, X[5] 3, W5, verb R3: getrest X[1] X[6] S2 intersectsort X[6] cons testinstlist X[6] 3, W7 unifyfirst X[6] X[4] testNILrest X[6] S3 R4: unifyfeature X[0] HEAD, X[3] unifyfeature X[0] SUBJECT, X[5] R5: jump W8 W0: pushcell X[0] setsort X[0] vp W1: pushcell X[1] S4 setfeature X[0] ARGS, X[1] setsort X[1] cons W2: pushcell X[2] setsort X[2] verb W3: pushcell X[3] setfeature X[2] HEAD, X[3] writetest 3, R1 W4: pushcell X[4] setfeature ....

[Article contains additional citation context not shown here]

H. Ait-Kaci and R. Nasr. Login: A logic programming language with built-in inheritance. Journal of Logic Programming, 3:185--215, 1986.

....1 u a 2 = b 2 u : u a n = b n u a 1 = a 2 u a 2 = a 3 u : u a n Gamma1 = a n then the original graph will have 2n 1 nodes, but 2n of these are collapsed by normalization step 8. To discover this efficiently, we use a version of A Kaci s algorithm for unifying Psi terms (Ait Kaci, 1984; Ait Kaci Nasr, 1986); the algorithm relies on the UNION FIND technique to identify nodes to be merged, and runs in time just slightly more than linear in the number of nodes in N . Therefore the cost of the non recursive portion of graph normalization is roughly linear in the number of nodes in it. The merging of ....

Ait-Kaci, H., & Nasr, R. (1986). LOGIN: A logic programming language with built-in inheritance. Journal of Logic Programming, 3, 187--215.

....of positive examples and background knowledge including type hierarchy (or the semantic attribute system) The output is a set of Horn clauses (Lloyd, 1987) having variables with type information. That is, the term is extended to the term. 4. 3 terms terms are the restricted form of terms (AitKaci and Nasr, 1986; Ait Kaci et al. 1994) Informally, terms are Prolog terms whose variables are replaced with variable Var of type T , which is denoted as Var:T. Predicate and function symbols are allowed to have features (or labels) For example, speak(agent)X:human,object)Y :language) is a clause based on ....

H. Ait-Kaci and R. Nasr, LOGIN: A logic programming language with built-in inheritance, Journal of Logic Programming, 3, pp.185-- 215, 1986.

....be bound either to an object or to another unbound variable Y. In the latter case, if Y is subsequently bound, its binding also becomes the binding of X. Since objects in this system are classified by their types, it is natural and useful to classify variables in a similar way. Thus, following [Ait kaci and Nasr 1985], a variable may be specified to be of a certain type. This is done by appending the name of a class to the variable name, using : as a separator, as in X:STUDENT. The consequence of assigning a type to a variable is that any binding of the variable is required to be a member of the variable s ....

....such a way that c 1 is a subclass of c 2 if and only if B(c 1 ) is (bit wise) less 18 than or equal to B(c 2 ) For a type t = fc 1 ; c k g, B(t) is constructed as the logical or of the B(c i ) The g.l.b. operation then reduces to logical and. The details of this scheme are described in [Ait kaci, et al. 1985]. 3 Assertions There are two main classifications of data: assertions and rules. Assertions, as described in this section, represent simple statements. Rules, which are used by the system to derive assertions from other assertions, are further classified as forward rules and backward rules. ....

Ait-kaci, H., and R. Nasr, LOGIN: A Logic Programming Language with Built-in Inheritance, MCC Technical Report AI-109-85.

....= 5 . 6 As can be observed, several singlevalued and multivalued properties are expressed as formulas consisting of a single atom (unit clauses) This logical representation differs from a more database oriented representation as sometimes employed in PROLOG like language [1]. In this case, the representation language is primarily viewed as a term manipulation language, not as a logical language. Then, the information above would be represented as a single term, for example as follows: patient(name = patient1 ; sex = female; age = 12; complaint = ....

H. Ait-Kaci and R. Nasr, LOGIN: a logic programming language with built-in inheritance. The Journal of Logic Programming, 3 (1986) 185-215.

....to give the metaterm an attribute which changes during the execution. This attribute can represent e.g. fl a list of goals that have to be executed when the variable becomes instantiated classical coroutining fl a goal to be executed to obtain or update the value of the variable fl a term [1] fl implementation of order sorted logic. This example generates a lazy list of 1 s. The tail of the list is initialised with a metaterm, whose attribute stores the goal ones 1. As soon as the list tail is unified, the handler procedure tail handler 2 drops the attribute, unifies the list tail ....

.... bind the metaterm to T call(Goal) and execute the goal : seterrorhandler(10, tailhandler 2) seterrorhandler(11, tailhandler 2) Example of use: addlist( X L] Y M] S R] S is X Y, addlist(L, M, R) addlist( 10 addlist( sepia] ones(Ones) addlist([1, 2, 3], Ones, Result) Ones = 1, 1, 1 M] Result = 2, 3, 4] More ( yes. 2.4.3 Generic Data Types Metaterms can represent a typed variable with run time type testing. The attribute can store the type name, type value, a list of possible type values or a goal that checks the type when the variable ....

[Article contains additional citation context not shown here]

Hassan A it-Kaci and Roger Nasr. LOGIN: A logic programming language with built-in inheritance. Journal of Logic Programming, 3(3):185--215, 1986.

....such as C logic and F logic (see [33] In these languages, the semantics of rule control is in the spirit of Prolog. These proposals have to be related to important efforts to provide formal foundations to object oriented programming such as for instance the works of Cardelli [20] or Ait Kaci [8]. For instance, the Psi terms of Login [8] are utilized in TEDM [40] an object oriented database system with deductive reasoning on objects. 9 3.7 Formal foundations of object oriented databases We have found few motivations so far that were specific to dooD. Most features that we discussed ....

....In these languages, the semantics of rule control is in the spirit of Prolog. These proposals have to be related to important efforts to provide formal foundations to object oriented programming such as for instance the works of Cardelli [20] or Ait Kaci [8] For instance, the Psi terms of Login [8] are utilized in TEDM [40] an object oriented database system with deductive reasoning on objects. 9 3.7 Formal foundations of object oriented databases We have found few motivations so far that were specific to dooD. Most features that we discussed were directly imported from the ooD approach ....

H. Ait-Kaci. Login: a logic programming language with built-in inheritance. J. Logic Programming, 3:185--215, 1986.

.... root are the unification grammar formalisms FUG [14] and LFG [13] developed for applications in computational linguistics (see [8] for a more recent paper in this area) Another, independent root is Ait Kaci s term calculus [1, 2] which is the basis of several constraint programming languages [3, 4, 5]. Smolka [20] gives a unified logical view of most earlier feature formalisms and studies an expressive feature constraint logic. Feature trees appeared only recently with the work on FT [7, 6] To our knowledge the notion of an arity constraint is new. Carpenter s [8] extensional types are ....

H. Ait-Kaci and R. Nasr. LOGIN: A logic programming language with built-in inheritance. The Journal of Logic Programming, 3:185--215, 1986. 14

....] is defined as: S ] ft : a j a = maxwt(t; S)g n ft : a j t = a = 0g: 6. 2 Logic Programming with Inheritance Ait Kaci and Nasr describe a system that extends conventional logic programming languages by incorporating subtyping and inheritance directly into the unification algorithm [1, 2]. In this case, the notions of instance and merge are defined with respect to a type semilattice that is userdefinable. Objects in D are partially ordered type structures called terms. These concepts are defined more formally in [1] for our purposes, it suffices to note that if the set of type ....

....vp(vp(saw, np(a, man) pp(with, np(a, telescope) corresponding to the reading John saw x with a telescope, where x = a man . 20 7. 2 Logic Programming with Polymorphism and Subtyping An earlier section discussed the application of our framework to logic programming with inheritance [1, 2]. There, it was assumed that the subtype relation over the set of type constructors is a (finite) semilattice. The theory of terms [1, 2] can be formulated in terms of order sorted logic [36] It turns out that order sorted unification, i.e. unification when some types may be subtypes of other ....

[Article contains additional citation context not shown here]

H. Ait-Kaci and R. Nasr, "LOGIN: A Logic Programming Language with Built-in Inheritance", J. Logic Programming vol 3 no. 3, Oct. 1986, pp. 185-215.

....is used in this paper is new but similar data models exist in the literature. In particular, the notions of alternation and construction appear as classification and aggregation in both Hull and Yap s Format Model [HY84] and Kuper and Vardi s LDM [KV84] Ait Kaci s feature structures [AN86] are also related to the Demeter kernel model. Our abstraction algorithms can be adapted to abstract feature structures from examples (this was actually the starting point of our investigation) 7 Conclusions We have sketched a fast algorithm for bottom up design of single inheritance class ....

H. Ait-Kaci and R. Nasr. Login: a logic programming language with built-in inheritance. Journal of Logic Programming, 3:185--215, 1986.

....i.e. grammars. 3. 2 Representation of Feature Structures The most convenient graphical representation of TFSs is through attribute value matrices (AVMs) However, to represent a (totally well typed) feature structure linearly we use a first order term like notation, based upon At Kaci s # terms [4, 5], where the type plays a similar role to that of a function symbol and the features are listed in a fixed order. Reentrancy is implied by attaching identical tags (of the form i , i # 0) to reentrant TFSs. A term is normal if all its types are tagged, and if the same tag appears more than once, ....

Hassan At-Kaci and Roger Nasr. LOGIN: a logic programming language with built-in inheritance. Journal of Logic Programming, 3:185--215, 1986.

....or KAP mentioned at the beginning of this section. The indepence property means among other things that negative constraints can essentially be handled through entailment simplification. One origin of FT is Ait Kaci s term calculus [1] which is at the heart of the programming language LOGIN [3] and further extended in the language LIFE [5] with functions over feature structures thanks to a generalization of the concept of residuation of Le Fun [4] Other precursors of FT are the feature descriptions found in so called unification grammars [13, 12] developed for natural language ....

....constraints; for instance, the constructor constraint x : A(y; z) can be expressed equivalently as Ax xf1; 2g x1y x2z, if we assume that A is a sort and the numbers 1; 2 are features. Second, it seems attractive to extend FT such that it can accommodate a sort lattice as used in [1, 3, 4, 5, 23]. One possibility to do this is to assume a partial order on sorts and replace sort constraints Ax with quasi sort constraints [A]x whose declarative semantics is given as [A]x j BA Bx: Given the assumption that the sort ordering has greatest lower bounds if lower bounds exist, it seems that ....

H. Ait-Kaci and R. Nasr. LOGIN: A logic programming language with built-in inheritance. The Journal of Logic Programming, 3:185--215, 1986.

....in two essential ways: ffl using a data structure richer than that provided by first order constructor terms; and, ffl allowing interpretable functional expressions as bona fide terms. The first extension is based on terms which are attributed partially ordered sorts denoting sets of objects [1, 3]. In particular, terms generalize first order constructor terms in their role as data structures in that they are endowed with a unification operation denoting type intersection. This gives an elegant means to incorporate a calculus of multiple inheritance into symbolic programming. Importantly, ....

Hassan At-Kaci and Roger Nasr. LOGIN: A logic programming language with built-in inheritance. Journal of Logic Programming, 3:185--215 (1986).

No context found.

Hassan At-Kaci. LOGIN: A Logic Programming Language With Built-In Inheritance. Tech. Rep., Microelectronics and Computer Technology Corporation, 1985.

No context found.

H. Ait-Kaci & R. Nasr, LOGIN: A Logic Programming Language with Built-in Inheritance, Journal of Logic Programming 3, 185--215, 1986.

No context found.

Hassan Ait-Kaci and Roger Nasr. LOGIN: A logic programming language with built-in inheritance. Journal of Logic Programming, 3(3):187--215, 1986.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC