Flener, P., and L. Popelmnsky, 1994, "On the Use of Inductive Reasoning in Program Synthesis: Prejudice and Prospects," in Logic Programming Synthesis and Transformation, MetaProgramming in Logic: Fourth International Workshops, LOBSTR'94 and META'94, Pisa, Italy, pp. 69-87, Springer-Verlag, Berlin.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....the quality of the example set in comparison to some of ILP programs. The quality of learning has been tested both on good examples and on randomly chosen example sets. 1 Top down learners Considering interactive generate and test top down learners in the context of automatic logic programming [5], four main drawbacks are being observed: 1. Too many positive examples are needed 2. The usefulness of the negative examples depends on the particular learning strategy 3. Generate (a hypothesis) and test (on the example set) strategy is too une#cient 4. Too many queries to the user are ....

.... to 46 seconds for append 3. 4.1.3 Discusion of results For predicates like last 2, delete 2 WiM is not capable to learn the correct target definitions from positive examples only. For all predicates above, we need at worst as many positive examples as CRUSTACEAN and less then FILP [3] see [5] for the comparison) with no need of negative examples. The number of queries to the user is less than for FILP (see [3] for more information) MIS is not able to learn from only positive examples. FOIL needs much more positive examples to succeed. Also the close world assumption employed by FOIL ....

Flener P., Popelnsk L.: On the use of inductive reasoning in program synthesis: Prejudice and prospects. Proc. of the 4th Int'l Workshop on Logic Program Synthesis and Transformation (LOPSTR'94), Pisa, Italy, 1994.

.... Knowing a concept means that one can act as a decision procedure for answering queries for that concept. But it doesn t necessarily imply the ability to actually write a decision procedure for that concept. Algorithmic ILP systems are ussualy assumed to be used only in program synthesis task. In [7] we suggest to use them in database technology or tutoring. In this paper we address the possibilities of algorithmic ILP methods in object oriented database schema modelling, i.e. in database schema design and restructuring. In deductive object oriented databases [1, 3, 17] both classes and ....

Flener P. and Popel#nsk# L.: On the use of inductive reasoning in program synthesis: Prejudice and prospects. Proc. of the 4th Int'l Workshop on Logic Program Synthesis and Transformation (LOPSTR'94), Pisa, Italy, 1994.

....solve the conflict between readability and preciseness of specifications, between the different conceptual worlds of the domain specialist and the software developer. We also offer an effective and viable solution to the formal informal controversy [Meyer 85; Deville 90; Wing 90; Fraser et al. 91; Flener Popelinsky 94] which combines the advantages of both informal and formal specifications. The executability of the internal representation and the semantics preserving mapping between the internal and external representations enables us to simulate the execution of the specification on the level of the external ....

.... number of publications and the existence of conferences specifically dedicated to logic program synthesis and transformation [Clement Lau 92; Lau Clement 93; Deville 94] Comprehensive overviews and pointers to the literature can be found in three recent publications [Bundy 92; Flener 93; Flener Popelinsky 94] Transformational implementation supports also the single document requirement. Especially, if the transformations can be performed automatically, the internal representation of the specification forms the only relevant document of the software. Instead of programs only simpler specifications ....

[Article contains additional citation context not shown here]

P. Flener, L. Popelinsky, On the Use of Inductive Reasoning in Program Synthesis: Prejudice and Prospects, in: Proceedings of LOPSTR '94, Pisa, -- 21 --

....February 1994. 7] Y. Deville and K. K. Lau. Logic program synthesis. The Journal of Logic Programming, 19 20:321 350, May 1994. 8] M. Ducass e and J. Noy e. Logic programming environments: Dynamic program analysis and debugging. The Journal of Logic Programming, 19 20:351 384, May 1994. [9] J. Jaffar and M. J. Maher. Constraint logic programming: A survey. The Journal of Logic Programming, 19 20:503 582, May 1994. 10] B. Le Charlier and P. Van Hentenryck. Experimental evaluation of a generic abstract interpretation algorithm for Prolog. ACM Transactions on Programming Languages ....

....Notes in Computer Science 844, pages 465 466. Springer Verlag, 1994. Poster Abstract. 8] D. A. Basin. Isawhelk: Whelk interpreted in Isabelle. In P. Van Hentenryck, editor, Proceedings of the Eleventh International Conference on Logic Programming, page 741. The MIT Press, 1994. Poster Abstract. [9] F. Bergadano and D. Gunetti. Inductive synthesis of logic programs and inductive logic programming. In Deville [30] pages 45 56. 10] A. Bossi and N. Cocco. Preserving universal termination through unfold fold. In G. Levi and M. Rodriguez Artalejo, editors, Proceedings of the Fourth ....

P. Flener and L. Popelinsky. On the use of inductive reasoning in program synthesis: prejudice and prospects. In Logic Program Synthesis and Transformation. Proceedings of LOPSTR'94, Pisa, Italy, 1994. To be published.

....February 1994. 7] Y. Deville and K. K. Lau. Logic program synthesis. The Journal of Logic Programming, 19 20:321 350, May 1994. 8] M. Ducass e and J. Noy e. Logic programming environments: Dynamic program analysis and debugging. The Journal of Logic Programming, 19 20:351 384, May 1994. [9] J. Jaffar and M. J. Maher. Constraint logic programming: A survey. The Journal of Logic Programming, 19 20:503 582, May 1994. 10] B. Le Charlier and P. Van Hentenryck. Experimental evaluation of a generic abstract interpretation algorithm for Prolog. ACM Transactions on Programming Languages ....

....Notes in Computer Science 844, pages 465 466. Springer Verlag, 1994. Poster Abstract. 8] D. A. Basin. Isawhelk: Whelk interpreted in Isabelle. In P. Van Hentenryck, editor, Proceedings of the Eleventh International Conference on Logic Programming, page 741. The MIT Press, 1994. Poster Abstract. [9] F. Bergadano and D. Gunetti. Inductive synthesis of logic programs and inductive logic programming. In Deville [30] pages 45 56. 10] A. Bossi and N. Cocco. Preserving universal termination through unfold fold. In G. Levi and M. Rodriguez Artalejo, editors, Proceedings of the Fourth ....

P. Flener and L. Popelinsky. On the use of inductive reasoning in program synthesis: prejudice and prospects. In Logic Program Synthesis and Transformation. Proceedings of LOPSTR'94, Pisa, Italy, 1994. To be published.

....we contemplate are algorithmically speci ed, the information generated either guidance for problem speci cation enhancement or an executable module is generated automatically. This is the essence of what induction algorithms have to contribute to the eld of automated software engineering [5, 12]. It has been noted [8] that the addition of input output examples to even complete speci cations may be bene cial (to human programmers) namely as a means of clari cation. The common fear that such examples may be inconsistent with the rest of the speci cation is unreasonable, as absence of ....

P. Flener and L. Popelnsky. On the use of inductive reasoning in program synthesis: Prejudice and prospects. In: L. Fribourg and F. Turini (eds), Proc. of LOPSTR/META'94, pp. 69-87. LNCS 883. Springer-Verlag, 1994.

.... translations of tracts of this thesis) The first author has successfully used these ideas in several medium sized projects [28] The second author has used them for debunking some of the myths on deductionbased and induction based approaches to the (semi )automatic synthesis of (logic) programs [11]. 2 The Role and Nature of Specifications In this section, we more closely examine specifications of programs. Such specifications are the essential pivot of the whole programming activity: without good specifications, it is impossible to understand what the correctness of a program means and ....

P. Flener and L. Popel'insk'y. On the use of inductive reasoning in program synthesis: Prejudice and prospects. In L. Fribourg and F. Turini (eds), Proc. of META'94 and LOPSTR'94, pp. 69--87. LNCS 883, Springer-Verlag, 1994.

....but, as mentioned earlier, this is rarely achieved in practice. The problem descriptions investigated here (the evidence) are actually even assumed to be incomplete. They are furthermore the most declarative (formal) descriptions that we can imagine (if they are constrained to be non recursive [16]) The latter description should give the predicate symbol representing the intended relation, the sequence of names and types of its formal parameters, pre conditions (if any) on these parameters, as well as the representation conventions of the formal parameters so that one knows how to ....

P. Flener and L. Popelínsky. On the use of inductive reasoning in program synthesis: Prejudice and prospects. In L. Fribourg and F. Turini (eds), Joint Proc. of META'94 and LOPSTR'94, pp. 69--87. LNCS 883, Springer-Verlag, 1994.

....right, and it gives rise to important applications. Recursive programs actually compute something, in the traditional understanding of what a program is and does, but such is not the case with all non recursive programs, which might for instance classify data as belonging to one concept or another [32]. Inferring recursive programs from assumed to be complete information such as the axiomatisation subset(S,L) X (member(X,S) member(X,L) where member is a known predicate (with the usual meaning) is called program synthesis, and features two main approaches, namely deductive synthesis ....

....the induction of generate and test programs could be efficiently and effectively guided. Just consider the potentially huge set of background knowledge predicates. This brings us directly to a first problem of many existing inductive synthesisers, namely their background knowledge usage bottleneck [32]. In a realistic programming scenario, the background knowledge consists of clauses for numerous predicates, just like with human programmers. However, we humans tend to dynamically organise this background knowledge according to relevance criteria, so that we do not think of using a definition of ....

[Article contains additional citation context not shown here]

P. Flener and L.Popelínsky. On the use of inductive reasoning in program synthesis: Prejudice and prospects. In L. Fribourg and F. Turini (eds), Joint Proc. of META'94 and LOPSTR'94, pp. 69--87. LNCS 883, Springer-Verlag, 1994.

....(and includes translations of tracts of this thesis) as well as on numerous discussions between the two authors. The second author has used some of these ideas for debunking some of the myths on deduction based and induction based approaches to the (semi )automatic synthesis of (logic) programs [9]. The remainder of this paper is organized as follows. In Section 2, we elaborate on our view of specifications. Based on the evidence that a program must be useful in some sense and must therefore have an understandable purpose, we argue that the specification of a program is precisely the link ....

P. Flener and L. Popel'insk'y. On the use of inductive reasoning in program synthesis: Prejudice and prospects. In L. Fribourg and F. Turini (eds), Proc. of META'94 and LOPSTR'94 , pp. 69--87. LNCS 883, Springer-Verlag, 1994.

....However, we realize that a cooperation between those streams is necessary to solve real problems of software engineering, and that this would be advantageous for both of them. For more information on the need for cross fertilization and cooperation between deductive and inductive synthesis, see [17]. At present, more attention is paid to the employment of ILP techniques in automatic programming [2, 4, 20, 8] The goal of the paper is to show what makes inductive synthesis of logic programs speci c, how it dioeers from concept learning as well as to analyze consequences of this dioeerence. It ....

Flener P. and Popel#nsk# L.: On the use of inductive reasoning in program synthesis: Prejudice and prospects. To appear in: Fribourg F. and Turini F. (eds): Proc. of LOPSTR'94 and META'94, Pisa. LNCS, Springer-Verlag, 1994.

.... on the application area (see Section 4) Recursive programs actually compute something, in the traditional understanding of what a program is and does, but such is not the case with all non recursive programs, which might for instance merely classify data as belonging to one concept or another [28]. Inferring recursive programs from assumed to be complete information such as the axiomatization subset(S,L) X (member(X,S) member(X,L) where member is a known predicate (with the usual meaning) is called program synthesis, and features two main approaches, namely deductive synthesis ....

....but, as mentioned earlier, this is rarely achieved in practice. The relation descriptions investigated here (the evidence) are actually even assumed to be incomplete. They are furthermore the most declarative (formal) descriptions that we can imagine (if they are constrained to be non recursive [28]) The latter description should give the predicate symbol representing the intended relation, the sequence of names and types of its formal parameters, pre conditions (if any) on these parameters, as well as the representation conventions of the formal parameters so that one knows how to ....

[Article contains additional citation context not shown here]

P. Flener and L.Popelínsky. On the use of inductive reasoning in program synthesis: Prejudice and prospects. In L. Fribourg and F. Turini (eds), Joint Proc. of META'94 and LOPSTR'94, pp. 69--87. LNCS 883, Springer-Verlag, 1994.

....They are themselves recursively defined, and may be called by general purpose learners for necessary or useful predicate invention. 4 Objectives of this Paper. I can now state my objectives more precisely: I will here focus on the process of useful predicate invention in a very relevant niche [13] of ILP, namely the learning of recursive logic programs. By computational concepts I will mean concepts that have (among others) recursive 1 logic programs as concept descriptions; and by (inductive) synthesis I will mean the learning of recursive logic programs for computational concepts. The ....

....that can only synthesize recursive logic programs for computational concepts. Examples of inductive synthesizers are the aforementioned SYNAPSE, ITOU, CILP, SIERES, and METAINDUCE. Their applicability as predicate inventors for general purpose learners further strengthens my former arguments [13] for their being a relevant niche of ILP, in addition to their relevance to (inductive) software engineering in general. In this paper, I will show that the knowledge (or mere conjecture) that there is a recursive program for the invented predicate gives us some extra leverage over conventional ....

Pierre Flener and Lubos Popelínsky. On the use of inductive reasoning in program synthesis: Prejudice and prospects. In L. Fribourg and F. Turini (eds), Proc. of META'94 and LOPSTR'94. LNCS nnn:xxx--yyy, Springer-Verlag, 1995.

....However, we realize that a cooperation between those streams is necessary to solve real problems of software engineering, and that this would be advantageous for both of them. For more information on the need for cross fertilization and cooperation between deductive and inductive synthesis, see [17]. At present, more attention is paid to the employment of ILP techniques in automatic programming [2, 4, 20, 8] The goal of the paper is to show what makes inductive synthesis of logic programs specific, how it differs from concept learning as well as to analyze consequences of this difference. ....

Flener P. and Popel'insk'y L.: On the use of inductive reasoning in program synthesis: Prejudice and prospects. To appear in: Fribourg F. and Turini F. (eds): Proc. of LOPSTR'94 and META'94, Pisa. LNCS, Springer-Verlag, 1994.

....on the quality of the example set in comparison to some of ILP programs. The quality of learning has been tested both on good examples and on randomly chosen example sets. 1 Top down learners Considering interactive generate and test top down learners in the context of automatic logic programming [5], four main drawbacks are being observed: 1. Too many positive examples are needed 2. The usefulness of the negative examples depends on the particular learning strategy 3. Generate (a hypothesis) and test (on the example set) strategy is too uneOEcient 4. Too many queries to the user are asked We ....

.... to 46 seconds for append 3. 4.1.3 Discusion of results For predicates like last 2, delete 2 WiM is not capable to learn the correct target definitions from positive examples only. For all predicates above, we need at worst as many positive examples as CRUSTACEAN and less then F ILP [3] see [5] for the comparison) with no need of negative examples. The number of queries to the user is less than for F ILP (see [3] for more information) MIS is not able to learn from only positive examples. FOIL needs much more positive examples to succeed. Also the close world assumption employed by ....

Flener P., Popel#nsk# L.: On the use of inductive reasoning in program synthesis: Prejudice and prospects. Proc. of the 4th Int'l Workshop on Logic Program Synthesis and Transformation (LOPSTR'94), Pisa, Italy, 1994.

No context found.

Flener, P., and L. Popelmnsky, 1994, "On the Use of Inductive Reasoning in Program Synthesis: Prejudice and Prospects," in Logic Programming Synthesis and Transformation, MetaProgramming in Logic: Fourth International Workshops, LOBSTR'94 and META'94, Pisa, Italy, pp. 69-87, Springer-Verlag, Berlin.

No context found.

Flener P. and Popel#nsk# L.: On the use of inductive reasoning in program synthesis: Prejudice and prospects. Proc. of the 4th Int'l Workshop on Logic Program Synthesis and Transformation (LOPSTR'94), Pisa, Italy, 1994.

No context found.

Flener P., Popelnsk L.: On the Use of Inductive Reasoning in Program Synthesis: Prejudice and Prospects. LOPSTR'94, Pisa, Italy.

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