Abiteboul, S., Vianu, V.: Procedural and Declarative Database Update Languages. In: Proc. 7th PODS. (1988) 240--250  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... and the set is ground and consistent (i.e. it does not contain complementary updates on the same fact) U Datalog was tailored to the need of keeping many of the already results and techniques developed for Datalog, specifically efficient query eval uation strategies [11] expressive power [5] and the nice collect execute model which nicely fits into a transaction model [53] The language we propose in this paper (called Obj U Datalog) is based on the notion of deductive object. Each object is an U Datalog database. Each object has a state (a set of facts) and a set of methods ....

....set is not consistent the trensection is aborted end no updete in the set is performed. The notion of consistency is en iraportent one, in thet it prevents e set of updetes conreining both en insertion end e deletion of the seine feet to be executed. By contrest in DLP [47] D [54] Tq [20] end DL [5], updetes ere executed es soon es they ere evelueted. This epproech leeds to complex sementics end to computetions performed in e sequence of stetes 12 instead of in a single one. In the following we recall the language and its informal behavior (see [53] for a complete description) We consider ....

[Article contains additional citation context not shown here]

S. Abiteboul and V. Vianu. Procedural and Declarative Database Update Languages. In Proc. of the Seventh A CM SIGA CT- SIGMOD-SIGART Syraposium on Principles of Database Systems, pages 240 251. ACM Press, 1988.

.... at least two di#erent topics: An update and a transaction language should be provided and typical atomicity, consistency, isolation, and durability (ACID) properties have to be guaranteed [24] This requirement has led to the definition of several languages integrating logic and updates [1,2,9,10,15,16,31,32,35]. In general, all those proposals are based on including special atoms denoting updates in (typically Datalog) rules. The proposed approaches di#er for several aspects, such as the semantics assigned to the resulting language (declarative vs. operational) the evaluation techniques, the update ....

S. Abiteboul and V. Vianu. Procedural and Declarative Database Update Languages. In Proc. of the Eighth ACM SIGACT-SIGMOD-SIGART Symp. on Principles of Database Systems, pages 240--250, 1988.

.... capabilities and can thus be viewed as an extension of relational database query languages, such as SQL [22] In particular, Datalog is a language of logic programming, based on Horn clauses without function symbols [22] Recently, Datalog has been extended to include negation (Datalog ) [2, 10], sets and lists [16] and functions [1] 3 In this paper we present a formalism for distributed knowledge bases based on modal logic [9] which has also been the basis for modal logics of knowledge and belief [7, 14] In modal logics of knowledge and belief a single agent (or more generally, ....

....modal logics of knowledge and belief. We define the model theoretic semantics of Datalog K via implication in a proper subset of first order modal logic tailored to Datalog K logic programs. We then define the operational semantics of Datalog K in terms of two inflationary fixpoint operators [2, 10], which are defined via the successive application of the one step INFER operator which returns all the possible facts that can be inferred from the EDDB by using the IDDB (which is a Datalog K logic program) The local fixpoint operator computes the inflationary fixpoint of the IDDB with ....

[Article contains additional citation context not shown here]

S. Abiteboul and V. Vianu, Procedural and declarative database update languages, Proceedings of ACM Symposium on Principles of Database Systems, Austin, Texas, 1988, pp. 240-250.

....1 ( u( OE (fhr 1 ; r 2 i; hs 1 ; s 2 ig) Transformation expressions will be denoted by ; 0 ; while Theta denotes the set of all transformation expressions. In the sequel we shall leave out extra parenthesis symbols wherever there is no risk of confusion. Abiteboul and Vianu [AV87, AV88] define a class of non deterministic transformations on databases that they call updates. This class is similar to our transformations in that it includes queries and modifications of the database state as special cases. They define an update as a relation between instances of a fixed schema s and ....

....is one of u or t. As we noted in Section 2, the transformations described by expressions in this class fall within the class of deterministic updates defined by Abiteboul and Vianu. It follows immediately that every query in ST is expressible in their languages detTL and detDL, which are shown in [AV88] to express all deterministic updates. It follows from Theorem 5.2 below that this inclusion is proper, since ST does not go beyond the second order queries SO. However, ST does include all the existential second order queries: let ST i denote the subclass of ST expressions that use at most i ....

S. Abiteboul & V. Vianu. Procedural and declarative database update languages. In: Proceedings of the Seventh ACM SIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems , pages 240--250, 1987.

....and update propagation. We model the sources of information as deductive databases, sharing the same logical language to express queries and updates, but containing independent, even if possibly related, data. Several approaches have been proposed integrating logic and updates (see for example [1,2,6,7,13,14,32,34,35,38,45]) The language we propose is based on the U Datalog language [6] and extends it with support for active rules and with the ability to model heterogeneous databases. U Datalog has been introduced with the aim of providing a set oriented logical update language, guaranteeing update parallelism in ....

S. Abiteboul and V. Vianu. Procedural and Declarative Database Update Languages. In Proc. of the Eighth ACM SIGACT-SIGMOD-SIGART Symp. on Principles of Database Systems, pages 240-250, 1988.

....of complexity theory. Recently parallel evaluation of recursive queries has also drawn considerable attention, CK85] AC89] In the past the emphasis has been to develop a natural set of primitives for a query language so that it can compute all the computable queries as in [CH80] Ch81] [AV88]. Unbounded arity relations or the ability to create new values have been used. For example, Chandra and Harel, in [CH80] define the concept of computable queries and present a complete database programming language and show that relational algebra augmented with the power of iteration is ....

....query language captures the primitive recursive properties. We then show that natural restrictions of this language capture P , DSPACE(log n) and NSPACE(log n) The expressive power of the bounded loop construct or its variant has been studied before in [AU79] Va82] Q89] CH80] Imm87] [AV88] but not in this framework. In [Ch81] Chandra raises the question of specifying a set of primitives of the form forall tuples t in relation R do statement S, where S is restricted so as not to use the order in which the forall cycles over all the tuples, such that programs in this style 1 Here a ....

[Article contains additional citation context not shown here]

S. Abiteboul and V. Vianu, Procedural and declarative database update languages. Proceedings of the Seventh ACM SIGACT-SIGMOD-SIGART Symposium on PODS, 1988, 240-250. 30

....Basic updates in a rule head are triggered, when the condition of the rule body is satis ed. There exist many di erent seman 31 tics depending on how the rules (set oriented tuple oriented, deterministically non deterministically, simultaneously successively, etc. are evaluated, see e.g. AV88] Some approaches like Statelog [LHL95] and the concept of Zaniolo [Zan93] de ne a clear semantics based on temporal logic programming [AM89] by incorporating states into the rule formalism. Lud ascher, May, and Lausen [LML96] present an extension of Statelog by update procedures and sequential ....

S. Abiteboul and V. Vianu. Procedural and declarative database update languages. In Proc. 7th ACM Symp. on Principles of Database Systems, pages 240-250, 1988.

....separates the query language from the update language. However, the most interesting languages and systems are those that integrate the query and update languages, for example, Dynamic logic programming (DLP) 41] LDL [43] transaction logic programming (T R ) 14] Declarative Languages (DL) [3], and RDL1 [20] Languages, providing integrated query and update facilities, can in turn be classified depending on whether updates are defined in the rule bodies (DLP, LDL and TR ) or in the rule heads (DL and RDL1) These two approaches were developed with different goals. The first family was ....

S. Abiteboul and V. Vianu. Procedural and Declarative Database Update Languages. In Proc. of the ACM Symposium on Principles of Database Systems, pp. 240--251. ACM, New York, USA, 1988.

....name is therefore itself. Where a naming function symbol is being used in a non naming way (for example it has a nonname as an argument) we make no assumption about its denotation. 5 Transaction Logic There have been several approaches to defining a logic of updates in recent years including [1, 2, 10, 22, 25]. One of the most fully developed is Transaction Logic [5, 7] by Bonner and Kifer. We briefly summarise (and simplify) the main features of Transaction Logic here in order to compare it with our approach. A theory of Transaction Logic (hereafter TR ) comprises: ffl the transaction base P , which ....

....k is a limit ordinal is similar. 8 RELATION TO OTHER WORK 19 8 Relation to Other Work Our approach to updates has its foundation in Transaction Logic. Bonner and Kifer s papers on Transaction Logic [6, 7] contain substantial reviews of related literature, including work on updating databases [2, 1, 25], the situation calculus [30] the event calculus [23, 22] and temporal logic [3] The paper by Reiter [30] contains some further remarks comparing Transaction Logic to the situation calculus. We invite the reader to consult these papers for comparison with related work. Instead of repeating ....

S. Abiteboul and V. Vianu. Procedural and declarative database update languages. In Symposium on Principles of Database Systems, pages 240-- 250, 1988.

....F LOGIC 84 Moshe Vardi, David Warren and Gio Wiederhold for discussions that helped shape this paper. Last, but not least, thanks goes to the anonymous referees for their constructive critique. A Appendix: A Perfect Model Semantics for F logic To adapt the various semantics for negation (such as [89, 46, 105, 104, 90, 5, 62]) to F logic, the general principle is to use method names in contexts where predicates are used in the classical setting. For the perfect model semantics [89] this means that stratification has to be ensured with respect to method names, so the program X [wants Y ] X [has Y ] would be ....

S. Abiteboul and V. Vianu. Procedural and declarative database update languages. In ACM SIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems (PODS), pages 240-- 250, 1988.

....forth. In [27] we use this representation of the natural numbers to prove that Hyperlog is computationally complete. In particular, we show that Hyperlog can simulate counter programs which are known to be computationally complete [17] In that paper we also show that Hyperlog is update complete [1]. 6 Implementation Figure 9 shows the architecture of our current Hyperlog implementation. The Database (DB) Browser Display Generator, Evaluator and Translator modules are implemented in the functional database programming language PFL [26] while the UI modules are implemented in Tcl Tk. Figure ....

....Work Our work relates to a number of areas, including database update languages, graph based data models and visual database languages, which we discuss in turn below. 7. 1 Database update languages Hyperlog programs have inflationary fixpoint semantics, similarly to languages such as detDL [1] and IQL [2] Most similar to Hyperlog is IQL, a textual, typed, rule based language whose programs can dynamically generate new object identifiers as a result of variables appearing in the head of a rule and not in the rule body. However, in Hyperlog new hypernode labels are generated only if ....

S.Abiteboul S.V.Vianu, "Procedural and declarative database update languages", Proc. ACM Symposium on Principles of Database Systems, Austin, Texas, 1988, pp 240-250.

....is also useful (Section 2.4) Phi OE is equivalent to : OE Omega : Transaction formulas combine simple transactions into complex ones. However, we also need a way to specify elementary changes to a database. One way to define such changes is to build them into the semantics as in [17, 22, 7, 1, 20]. The problem with this approach is that adding new kinds of elementary transitions leads to a redefinition of the very notion of a model and thus to a revamping of the entire theory, including proofs of soundness and completeness. This drawback is quite serious, as there appears to be no small, ....

....for defining composite actions, and there is no facility, like T R s transition base, for defining elementary actions. Second, state changes in Process Logic are hypothetical, and there seems to be no easy to make them permanent. Abiteboul and Vianu developed a family of update languages [1], and provided impressive results on complexity and expressibility. However, these languages lack several features that are present in T R. First, they apply only to relational databases, not to arbitrary sets of first order formulas. Second, there is no facility for constraining transaction ....

[Article contains additional citation context not shown here]

S. Abiteboul and V. Vianu. Procedural and declarative database update languages. In ACM SIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems (PODS), pages 240--250, New York, 1988. ACM.

....for :OE, for any transaction formulas and OE. 3.2 Elementary Transitions Transaction formulas combine simple transactions into complex ones. However, we also need a way to specify elementary changes to the database. One way to define such transitions is to build them into the semantics as in [40, 48, 8, 18, 1, 43]. The problem with this approach is that adding new kinds of elementary transitions leads to a redefinition of the very notion of a model and thus to a revamping of the entire theory, including the need to reprove soundness and completeness results. In other words, such theories are not ....

....principles than T R s. In particular, it does not embrace the concept of the execution path, which is fundamental to T R. Consequently, dynamic constraints of the kind discussed in Section 2.6 and in [11] cannot be expressed. Abiteboul and Vianu developed a family of declarative update languages [1], including impressive results on complexity and expressibility. However, these languages lack several features that are present in T R. First, they apply only to relational databases, not to arbitrary sets of first order formulas. Thus, it is not possible to insert or delete rules from a ....

[Article contains additional citation context not shown here]

S. Abiteboul and V. Vianu. Procedural and declarative database update languages. In ACM SIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems (PODS), pages 240-- 250, 1988.

....what can be viewed as meta or higher order data. It is easy to see that every data definition that consists of definite Horn rules has a unique least Herbrand model, which can be taken as its semantics. With negative literals in rule bodies, there is a number of different competing semantics [GRS88, GL88, Van89, AV88, KP88], each of which can easily be adapted to T PC. All of these semantics have in common that the logical entailment is defined with respect to some subset of models, called canonic models (which may be different for different proposals) We can thus assume that there is a subclass of canonic models ....

S. Abiteboul and V. Vianu. Procedural and declarative database update languages. In Proceedings of the ACM SIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems, pages 240--250, 1988.

....transaction base) is equivalent to a conjunction of all the rules in the set. A transaction base defines complex formulas in terms of simpler ones. However, we also need a way to specify elementary changes to a database. One way to define such transitions is to build them into the semantics as in [17, 18, 9, 1]. The main problem with this approach is that adding new kinds of elementary transitions requires redefining the very notion of a model and, hence, entails an overhaul of the entire proof theory. This is a serious drawback since there appears to be no small, single set of elementary transitions ....

S. Abiteboul and V. Vianu. Procedural and declarative database update languages. In ACM SIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems (PODS), pages 240--250, New York, 1988. ACM.

....concepts but do not support relations naturally. Hence, recent research has aimed at integrating the two paradigms. The integration has generally taken the route of extending logic based deductive database languages with features such as object identity, sets, functions, methods and inheritance [1, 2, 3, 14]. In contrast, in this paper we report upon a graph based approach to such an integration. Our use of graphs has two key advantages : firstly, graphs are formally defined, well understood structures; secondly, it is widely accepted that graph based formalisms considerably enhance the usability of ....

.... ) CTR = 0 , SEQ = STEP2 , We conclude this section by examining the expressiveness of Hyperlog with respect to database updates. We first define what an update is in our context and then define the concept of update completeness, by analogy to previous work in relational databases [1, 8]. Given a type repository TR, we define the set inst(TR) to contain all hypernode repositories which are well typed with respect to TR. We define an update to be a partial recursive mapping from inst(TR) to inst(TR) that is C generic. Cgenericity was introduced in [21] and intuitively means that, ....

[Article contains additional citation context not shown here]

Abiteboul S. and Vianu V. Procedural and declarative database update languages. Proceedings of ACM Symposium on Principles of Database Systems, Austin, Texas (1988), 240-250.

.... logarithmic space and constant time, or equivalently, a uniform sequence of polynomial size, unbounded fan in boolean circuits of constant depth) The need for an order relation is a technicality to capture classes below NPTIME, which can be substituted by the use of non determinism [AV87, AV88] The property of a query language capturing precisely the set of queries expressible in a complexity class is central to the area of descriptive complexity [Imm88a] However, the notion of a language capturing a complexity class has not only theoretical relevance. In the author s opinion, ....

Serge Abiteboul and Victor Vianu. Procedural and declarative database update languages. In Proceedings of the Seventh ACM SIGACT-SIGMOD Symposium on Principles of Database Systems, pages 240--250. Assoc. for Comp. Machinery, 1988.

....the complexity of answering a fixed query in terms of the size of the database. The rationale behind data complexity is that the size of the database dominates the query size by several orders of magnitude for most applications. It is well known that any Datalog and any inflationary Datalog : [1, 21] query evaluated on traditional relational databases have PTIME data complexity. In [17] it is also shown that any inflationary Datalog : with rational order and any inflationary Datalog : with equality over an infinite set of constants query can be also evaluated in PTIME data complexity. The ....

S. Abiteboul, V. Vianu. Procedural and Declarative Database Update Languages. Proc. 7th ACM Symposium on Principles of Database Systems, 240-- 250, 1988.

....In T R, such a set of transaction formulas is called a transaction base. A transaction base defines complex formulas in terms of simpler ones. However, we also need a way to specify elementary changes to a database. One way to define such transitions is to build them into the semantics as in [25, 27, 16, 2]. A problem with this approach is that adding new kinds of elementary transitions leads to a redefinition of the very notion of a model and thus to an overhaul of the entire proof theory. This is a serious drawback since there appears to be no small, single set of elementary transitions that is ....

....lingua franca as SQL or QUEL. For example, inserting a set of tuples into a relation is a basic SQL operation, as is deleting a set of tuples from a relation. Yet, bulk updates like these are conspicuously absent from most logic based proposals for updating logic programs. The few exceptions are [13, 16, 27, 2], which are discussed in Section 4. The unusual difficulty with this kind of update seems to arise because most logical formulations of updates are based on the insertion and deletion of single tuples. This is not how SQL works, however. SQL first computes a query and then inserts the resulting ....

[Article contains additional citation context not shown here]

S. Abiteboul and V. Vianu. Procedural and declarative database update languages. In ACM SIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems (PODS), pages 240--250, New York, 1988. ACM.

....next by examining the actual state. Although bulk updates and sequential composition can be handled better than in the other approaches, there currently exist no concepts for modularization and integration of transaction models. Some other approaches dealing with updates in deductive databases are [1, 12, 20, 21]. 1] and [12] can be regarded as the starting point for the concepts mentioned in the previous paragraph. Reiter [20] characterizes similar active behaviour by the situation calculus. In [21] a modal logic for verifying database updates is developed. However, the updates are programmed in a ....

....the actual state. Although bulk updates and sequential composition can be handled better than in the other approaches, there currently exist no concepts for modularization and integration of transaction models. Some other approaches dealing with updates in deductive databases are [1, 12, 20, 21] [1] and [12] can be regarded as the starting point for the concepts mentioned in the previous paragraph. Reiter [20] characterizes similar active behaviour by the situation calculus. In [21] a modal logic for verifying database updates is developed. However, the updates are programmed in a language ....

S. Abiteboul and V. Vianu. Procedural and declarative database update languages. In Proc. 7th ACM SIGACT-SIGMOD-SIGART Symp. on Principles of Database Systems, pages 240--250, 1988.

....infinite set of arity k tuples (or points in k dimensional space D k ) ffl The syntax of a CQL is the union of an existing database query language and a decidable logical theory. For example: Relational calculus [15] the theory of real closed fields [53] Section 2) Inflationary Datalog : [1, 22, 33] the theory of dense linear order with constants (Section 3) Inflationary Datalog : the theory of equality on an infinite domain with constants (Section 4) and Datalog boolean equations (Section 5) In each of these cases, we combine in the obvious way the syntax of the database language ....

....x 2 g. The result (interpreting the generalized relation as an infinite set of points) of 9x:R(x; y) is the set fyjy 0g, which cannot be represented by polynomial equality constraints. 2 Datalog constraints: We now consider Datalog with constraints. The syntax is that of Datalog (e.g. see [1, 29, 33, 54, 55]) but we allow the bodies of rules to contain constraints. Definition 1.10 Let Phi be a class of constraints. Let R 1 ; R i ; be predicate symbols, each with a fixed arity. A Datalog Phi query program is a finite set of rules of the form: t 0 : t 1 ; t 2 ; t l : t 0 ....

[Article contains additional citation context not shown here]

S. Abiteboul, V. Vianu. Procedural and Declarative Database Update Languages. Proc. 7th ACM PODS, 240--250, 1988.

....but a positive fragment can be defined that only expresses eventually computable queries. The Datalog : languages yield some surprises: the standard semantics, stratified and well founded [51] are ill suited for expressing eventually computable queries, whereas the inflationary semantics [8, 69] turns out to be naturally suited to express such queries, and thus has an advantage over the first two semantics [10] 3.2 Query Languages The query languages proposed in the context of the Web vary depending on the target data. Some languages are aimed at querying the Web as a whole, based on ....

S. Abiteboul and V. Vianu. Procedural and declarative database update languages. In Proc. ACM PODS, pages 240--250, 1988.

....properties of deterministic and nondeterministic extensions of Datalog Serge Abiteboul and Eric Simon I.N.R.I.A. BP. 105, 78153 Le Chesnay December 30, 1993 Abstract Fundamental properties of deterministic and nondeterministic extensions of Datalog from [AV88] are studied. The extensions involve the use of negative literals both in bodies and heads of rules. Negative literals in heads are interpreted as deletions. A deterministic semantics is obtained by firing in parallel all applicable rules. The nondeterministic semantics results from firing ....

.... optimization [BR86] Recently, many proposals emerged to develop extensions of Datalog with increased expressive power, providing forms of nonmonotonic reasoning (see, for instance [Kan88, CH80, Apt87, BH86, GS88] The focus of the present paper is the study of extensions of Datalog proposed in [AV88, AV89]. These extensions form the basis of implementation efforts [dMS88] for so called production rule systems. The price to pay for the increased power is that nice properties of Datalog are lost, such as the existence of a least fixpoint, and the guarantee of program termination. In the present ....

[Article contains additional citation context not shown here]

S. Abiteboul and V. Vianu. Procedural and Declarative Database Update Language. In 7th ACM Symposium on Principles of Database Systems, pages 240--250, March 1988.

....saw that a rule such as R(a; b) is interpreted as an insertion if R is a base relation. One can insert new facts concerning a method in the same manner. Insertion in a class is more involved since we are inserting oids and not values. Object creation is controlled following the spirit of [8, 5]. We allow the presence of variables occurring in the head of a rule and not in the body assuming that they are class variables. These variables are interpreted as new objects. Continuing with the Bill of Material example. Suppose that we are given a set partnames of names and we want to have a ....

S. Abiteboul and V. Vianu. Procedural and Declarative Database Update Language. In Proc. ACM PODS, 240--250, 1988.

No context found.

Abiteboul, S., Vianu, V.: Procedural and Declarative Database Update Languages. In: Proc. 7th PODS. (1988) 240--250

First 50 documents

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