Francois Bancilhon. Naive evaluation of recursively defined relations. In Brodie and Mylopoulos, editors, On Knowledge Base Management Systems --- Integrating Database and AI Systems. Springer-Verlag, 1985.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....databases [22] and an important way station toward future knowledge management systems, which will have to combine efficient inference mecha nisms from logic with efficient and secure management of large information banks from database systems. Toward this goal, the 53 project, which began in 1984, has produced a new language, new techniques for compilation and query optimization, and an efficient and portable prototype. This paper recounts the experience and various lessons learned in this effort. A. Overview I. INTRODUCTION HE objective of the Logic Data Language ( D ) sys tem is to ....

.... of complex data intensive applications, and bridging the gap between the database and the programming language this gap is often described as an impedance mismatch [ 18] Yet, the approach favored by previous researchers has been that of interfacing relational DBMS s to traditional languages [45] 50] More recently, major efforts have been made to integrate databases and programming lan guages under the object oriented paradigm [28] These approaches tend to abandon relational databases in favor of object oriented ones often supporting a limited query capability and the navigational ....

[Article contains additional citation context not shown here]

F. Bancilhon, "Naive evaluation of recursively defined relations," in On Knowledge Base Management Systems, M. Brodie and J. MyIopoulos, Eds. New York: Springer-Verlag, 1985.

....be reached from the node a in the base graph. The recursion depth of a relation is defined as the length of the longest path in the base graph . 2. 2 Primitive Solution for TC Problem The simplest and most direct algorithm for evaluating a transitive closure of a relation is the Naive Evaluation [Ba86, BMSU86]. In Naive Evaluation, the rules are compiled into an iterative program in which a composition (that is, a join with a subsequent projection) is carried out to generate new tuples. They are then added to the result relation CLOSURE, which has previously been initiated with BASIS. The process is ....

....of Naive Evaluation and analyze them, in order to point out some improvements that have influenced the development of our efficient strategy. Avoidance of redundant workRedundant work occurs when rules are evaluated more than once based on the same data, i.e. operations are repeated senselessly [Ba86, BR86]. Primarily, there are two possible causes for it. One is the existence of cyclic data that directs the evaluation back to its start point (this will be discussed in some detail later) The other lies in the awkwardness of an algorithm. This is just the case of Naive Evaluation, since the ....

[Article contains additional citation context not shown here]

Bancilhon, F.: Naive Evaluation of Recursively Defined Relations, in: On Knowledge Base Management Systems - Integrating Database and AI Systems, (Ed: Brodie and Mylopoulos), Springer-Verlag, 1986, pp. 165-178.

....2. A Survey on View Maintenance: Applications and Techniques 21 terialized form. Another improvement of this method is that it does not require a special evaluation procedure for its implementation, but it can, quite naturally, be used with the standard naive and semi naive evaluation procedures [Ban85, Ull88] An optimization similar to magic sets [BR87, MPR90] is also incorporated into the algorithm. Mumick and Pirahesh [MP94] discuss the importance of integrating magic sets with traditional optimizations, such as selection pushing. 2.6 Other View Related Work 1. Queries Independent of ....

F. Bancilhon. Naive Evaluation of Recursively Defined Relations. In M.L. Brodie and J. Mylopoulos, editors, On Knowledge Base Management Systems: Integrating Artificial Intelligence and Database Technologies. Springer-Verlag, 1985. 115 BIBLIOGRAPHY 116

....perform better than re evaluation, at least in the majority of cases. Little analysis or experimentation, however, supports this belief. To the best of our knowledge, incremental methods for query computation have not been implemented (with the exception of semi naive evaluation of logic programs [Ban85] which also utilizes differential expressions to reduce the time of the computation) Incremental methods usually need to store much information in addition to query answers (i.e. many intermediate results) However, it is not clear that maintaining these partial results is better than ....

F. Bancilhon. Naive Evaluation of Recursively Defined Relations. In M.L. Brodie and J. Mylopoulos, editors, On Knowledge Base Management Systems: Integrating Artificial Intelligence and Database Technologies. Springer-Verlag, 1985.

....negation. 2 Suppose we have facts query(p(a; X) q(b; X) r(a; c) and r(a; d) Using rule R1 we can infer the facts q(b; c) and q(b; d) The use of this higher order syntax is not essential for our discussion, but it makes the presentation concise. We assume that Semi Naive evaluation [BR87a, Ban85] is used. In a bottom up evaluation, a derivation is made as follows: previously derived facts are unified with each predicate in the body of a rule; the head of the instantiated rule is the fact that is derived. Semi naive bottomup evaluation has the property that no derivation step is repeated ....

....a more accurate comparison, the cost of each inference has to be taken into account. 6 A Model for Bottom up Evaluation Our model for bottom up evaluation is as follows. The given program is first rewritten using Magic Templates with Tail Recursion (Section 4) and then by Semi Naive rewriting [BR87a, Ban85]. Next the rewritten program is evaluated using Semi Naive evaluation. We model the entire set of actions in a Semi Naive bottom up evaluation as a sequence of uses of facts to derive other facts. In an iteration, a fact r(a) is said to be newly derived if it was derived for the first time in the ....

Francois Bancilhon. Naive evaluation of recursively defined relations. In Brodie and Mylopoulos, editors, On Knowledge Base Management Systems --- Integrating Database and AI Systems. Springer-Verlag, 1985.

....U p 1 p 2 L (6.1 By a reasoning similar to Lemma 2.1, an upper bound on distance between p 3 and p 2 can be obtained as: p 3 p 2 U = c 3 c 2 c 3 p 3 p 2 c 2 We will now incorporate the above bounding procedure in a breadth first search [26] procedure. Note that the seminaive algorithm [6] also performs a breadth first search for determining reachability from a specified node. In the following algorithm, OPEN is a queue, each element of which is a tuple of the form node, distance where the distance field contains the best (largest) known distance from source to the corresponding ....

F. Bancilhon, "Naive Evaluation of Recursively Defined Relations," in On Knowledge Base Management Systems --- Integrating Database and AI Systems, M. Brodie and J. Mysopoulos (ed.), Springer-Verlag, 1985.

....the time cost of the evaluation shown is O(V) regardless of the sizes of the difference lists, ignoring the cost of setting up the initial query and printing the answer. No occur checks are performed. 2 4. 3 Correctness and Cost of Apply Rule We call a version of Semi Naive evaluation ([Ban85, BR87a]) that uses procedure Apply Rule to perform rule application as Opt NG SN evaluation. We call the query evaluation technique that first rewrites the program and query using MGU MTTR rewriting, and then evaluates it using Opt NG SN evaluation as Opt NGBU evaluation. 12 Theorem 4.1 Let P be a ....

Francois Bancilhon. Naive evaluation of recursively defined relations. In Brodie and Mylopoulos, editors, On Knowledge Base Management Systems --- Integrating Database and AI Systems. Springer-Verlag, 1985.

....this transitive closure e#ciently. The original query condition is now applied directly to the datum parent relation and not the derived anc relation, i.e. selection has been pushed inside recursion. Furthermore, a refinement of fixpoint known as semi naive fixpoint is used to solve this problem [Ban, BaR, Ull, SaZ4]. The seminaive fixpoint iteration basically will begin by computing the parents of marc and then the parents of the parents, and so on until no new ancestor is found. More complex rewriting is required, however, before the following query can be mapped into a single fixpoint: anc(X, brian) ....

Bancilhon, F., "Naive Evaluation of Recursively defined Relations", On Knowledge Base Management Systems, (M. Brodie and J. Mylopoulos, eds.), Springer--Verlag, 1985.

....nodes. 1. INTRODUCTION Transitive closure is regarded to be an important operation for the next generation of database systems [1, 5, 6, 12, 13, 15, 17, 19, 21] and considerable research has been devoted to designing algorithms for computing the transitive closure of database relations [2, 4, 9 11, 16, 24]. These algorithms can be classified into three major families. Iterative algorithms, such as semi naive [4] logarithmic [10, 24] and variations thereof [9, 10, 16] compute transitive closure by repeatedly computing a relational algebraic expression, stopping when no more new answer tuples are ....

....systems [1, 5, 6, 12, 13, 15, 17, 19, 21] and considerable research has been devoted to designing algorithms for computing the transitive closure of database relations [2, 4, 9 11, 16, 24] These algorithms can be classified into three major families. Iterative algorithms, such as semi naive [4], logarithmic [10, 24] and variations thereof [9, 10, 16] compute transitive closure by repeatedly computing a relational algebraic expression, stopping when no more new answer tuples are generated, after a number of iterations that depends on the underlying database. Direct algorithms, on the ....

F. Bancilhon, "Naive Evaluation of Recursively Defined Relations," in On Knowledge Base Management Systems --- Integrating Database and AI Systems, M. Brodie and J. Mysopoulos (ed.), Springer-Verlag, 1985.

....the literature that focus on minimizing the I O costs of path computation in a database setting that assumes a fixed size main memory I O buffer. Most of such research has proposed solutions to solve recursive query problems for general databases that focused on pure transitive closure computation [1, 4, 7, 13, 27, 28, 29, 35]. In our work, rather than aiming for generality, we now take an application driven stance by proposing different disk page clustering algorithms for optimizing path query processing for GIS type of applications and then experimentally evaluating their relative advantages and disadvantages. Two ....

Bancilhon, F., "Naive Evaluation of Recursively Defined Relations", 1985. In On Knowledge Base Management Systems - Integrating Database and AI systems, M. Brodie and J, Mylopoulos, Eds., Springer-Verlag, New York.

....facts that have been newly generated (and not yet used in inferences) The second component is a technique to apply the rewritten rules and update these differentials, ensuring that all derivations are made exactly once. Semi naive algorithms have been proposed by several researchers (e.g. [B85, BR87]) These algorithms evaluate the fixpoint in an iterative fashion, with every rule applied once in each iteration. In these algorithms, facts generated in an iteration can be used to generate other facts only in subsequent iterations. We present two fixpoint evaluation algorithms, General ....

....only in subsequent iterations. We present two fixpoint evaluation algorithms, General Semi Naive (GSN) and Ordered Semi Naive (OSN) the latter has a simple variant called Predicate Wise Semi Naive (PSN) These algorithms can use any of the semi naive rewriting techniques proposed earlier (e.g. [B85, BR87]) with minor modifications. GSN applies a rule to produce new facts, and then immediately makes these facts available to subsequent applications of other rules (possibly in the same iteration) PSN makes facts generated for a predicate p available after all rules defining p have been applied. ....

[Article contains additional citation context not shown here]

F. Bancilhon, "Naive Evaluation of Recursively Defined Relations," TR DB-004-85, MCC, 1985.

....transitive closure efficiently. The original query condition is now applied directly to the datum parent relation and not the derived anc relation, i.e. selection has been pushed inside recursion. Furthermore, a refinement of fixpoint known as semi naive fixpoint is used to solve this problem [Ban, BaR, Ull, SaZ4]. The seminaive fixpoint iteration basically will begin by computing the parents of marc and then the parents of the parents, and so on until no new ancestor is found. More complex rewriting is required, however, before the following query can be mapped into a single fixpoint: anc(X, brian) ....

Bancilhon, F., "Naive Evaluation of Recursively defined Relations", On Knowledge Base Management Systems, (M. Brodie and J. Mylopoulos, eds.), Springer--Verlag, 1985.

....into a predicate connection graph [71] which is for storing the relationship between terms and the clause heads that can potentially be unified. The structure is also used to maintain the entry points for queries. Recursive rules are compiled by means of naive evaluation and magic sets methods [7, 8]. Then for a given query, the system generates all possible proof plans in the form of proof schema and compiles a single relational algebra program (RAP) which, when executed, produces the set of all possible answers to the query. The RAP is further optimized using relational database techniques ....

F. Bancilhon. Naive evaluation of recursively defined relations. Technical Report DB-004-85, Microelectronics and Computer Technology Corporation, 1985.

....queries [5] Datalog provides a way of incorporating recursion into a query language. However, it also raises the complexity of query evaluation. The problem of efficiently computing Datalog queries has attracted a great deal of attention in the database and logic programming communities e.g. [6, 8, 12, 14, 17, 18, 24, 25, 27]. In this paper, we consider the problem of repeatedly evaluating the same (computationally expensive) Datalog query to a database that is being updated between successive query requests. In this case, it should be possible to use the difference between successive database states and the answer to ....

....from four to one, and reduces the number of tuples accessed in the joins. This technique also applies to other examples described below. We will discuss the complexity of more general IEC in Section 3. 2 Example 2. 2 We now compare our incremental method with the semi naive evaluation method [6, 31], by considering their computations of the transitive closure query Q after edge(3; 4) is added to the database D in Example 2.1. To make the comparison fair for the semi naive method, we assume that the semi naive method also starts with path o available. We use D to denote the relation ....

[Article contains additional citation context not shown here]

F. Bancilhon. Naive evaluation of recursively defined relations. In M. L. Brodie and J. Mylopoulos, editors, On Knowledge Base Management Systems: Integrating Artificial Intelligence and Database Technologies. Springer-Verlag, 1985.

.... transformation [7] else use supplementary magic sets transformation [19] The compiler then chooses one of three strategies to evaluate the transformed program, according to the following decision procedure: if the transformed program is negation free then use semi naive bottom up evaluation [2]; else if the transformed program is stratified then use Kerisit Pugin s method [9] else use alternating fixpoint for magic programs [10] The Nail compiler makes these selections at compiler time, because all conditions in the above algorithms can be tested by checking only the program ....

F. Bancilhon. Naive evaluation of recursively defined relations. In M. L. Brodie and J. Mylopoulos, editors, On Knowledge Base Management Systems, pages 165-- 178. Springer-Verlag, New York, New York, 1986.

....queries [5] Datalog provides a way of incorporating recursion into a query language. However, it also raises the complexity of query evaluation. The problem of efficiently computing Datalog queries has attracted a great deal of attention in the database and logic programming communities e.g. [6, 8, 12, 14, 16, 17, 23, 24, 26]. In this paper, we consider the problem of repeatedly evaluating the same (computationally expensive) Datalog query to a database that is being updated between successive query requests. In this case, it should be possible to use the difference between successive database states and the answer to ....

....transformation of the original program into a more efficient nonrecursive program. The second basis is designed to fit database application by virtue of efficiency and easy programmability in database query languages. We now briefly compare our approach with related work. Semi naive evaluation [6]. The basic idea of semi naive evaluation is, in each iteration in the bottomup evaluation, to compute only those facts that depend on at least one fact computed in the previous iteration. This approach differs from our approach in three ways: i) the evaluation is incremental between iterations ....

F. Bancilhon. Naive evaluation of recursively defined relations. In M. L. Brodie and J. Mylopoulos, editors, On Knowledge Base Management Systems: Integrating Artificial Intelligence and Database Technologies. Springer-Verlag, 1985.

....see if the fixpoint has been reached. This amounts to checking if any new facts were computed in an iteration. Such a check also enhances efficiency in that subsumed facts can be discarded, and not be used to make any further derivations in subsequent iterations, if we use Semi naive evaluation ([Ban85, BR87]) Facts in a constraint query language (referred to as constraint facts) are conjunctions of constraints. Relations are finite collections of facts, as is usual in database query languages. In this paper, we concern ourselves with the problem of subsumption in constraint query languages where the ....

Francois Bancilhon. Naive evaluation of recursively defined relations. In Brodie and Mylopoulos, editors, On Knowledge Base Management Systems --- Integrating Database and AI Systems. Springer-Verlag, 1985.

....on centralized a route guidance system where TMCs can possess large main memory and high computation power, making the main memory path view approach feasible. Recently, there has been a body of literature on transitive closure computation and recursive query processing in the database community [1, 3, 4, 5, 6, 13, 14, 15]. While their works focus on secondary storage solutions, the benchmark studies in [3, 15] showed that the I O in computing all pair shortest paths becomes intolerable when the underlying graphs are cyclic with more than 300 nodes. If Shekar et al. s finding in [19] demonstrates that the current ....

F. Bancilhon, "Naive Evaluation of Recursively Defined Relations", In On Knowledge Base Management Systems -- Integrating Database and AI systems, M. Brodie and J, Mylopoulos, Eds., Springer-Verlag, New York, 1985

No context found.

Francois Bancilhon. Naive evaluation of recursively defined relations. In Brodie and Mylopoulos, editors, On Knowledge Base Management Systems --- Integrating Database and AI Systems. Springer-Verlag, 1985.

No context found.

Bancilhon, F. Naive evaluation of recursively defined relations. In Brodie and Mylopoulos, editors, On Knowledge Base Management Systems --- Integrating Database and AI Systems. Springer-Verlag (1985) 165--178.

No context found.

F. Bancilhon. Naive Evaluation of Recursively Defined Relations. On Knowledge Base Management Systems: Integrating AI and DB Technologies, 1986.

No context found.

Bancilhon, F. (1985). Naive evaluation of recursively defined relations. On Knowledge Base Management Systems (Islamorada) (pp. 165-178).

No context found.

Bancilhon, F., "Naive Evaluation of Recursively defined Relations", Unpublished Manuscript, 1985.

No context found.

Bancilhon, F., "Naive Evaluation of Recursively Defined Relations," TR. DB-004-85, MCC, Austin, Texas, 1985.

No context found.

Bancilhon, F., "Naive Evaluation of Recursively Defined Relations," TR. DB-004-85, MCC, Austin, Texas, 1985.

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