## Conjunctive-Query Containment and Constraint Satisfaction (1998)

### Cached

### Download Links

- [www.cse.ucsc.edu]
- [www.cs.rice.edu]
- [www.cs.rice.edu]
- DBLP

### Other Repositories/Bibliography

Venue: | Journal of Computer and System Sciences |

Citations: | 141 - 15 self |

### BibTeX

@INPROCEEDINGS{Kolaitis98conjunctive-querycontainment,

author = {Phokion G. Kolaitis and Moshe Y. Vardi},

title = {Conjunctive-Query Containment and Constraint Satisfaction},

booktitle = {Journal of Computer and System Sciences},

year = {1998},

pages = {205--213}

}

### Years of Citing Articles

### OpenURL

### Abstract

Conjunctive-query containment is recognized as a fundamental problem in database query evaluation and optimization. At the same time, constraint satisfaction is recognized as a fundamental problem in artificial intelligence. What do conjunctive-query containment and constraint satisfaction have in common? Our main conceptual contribution in this paper is to point out that, despite their very different formulation, conjunctive-query containment and constraint satisfaction are essentially the same problem. The reason is that they can be recast as the following fundamental algebraic problem: given two finite relational structures A and B, is there a homomorphism h : A ! B? As formulated above, the homomorphism problem is uniform in the sense that both relational structures A and B are part of the input. By fixing the structure B, one obtains the following non-uniform problem: given a finite relational structure A, is there a homomorphism h : A ! B? In general, non-uniform tractability results do not uniformize. Thus, it is natural to ask: which tractable cases of non-uniform tractability results for constraint satisfaction and conjunctive-query containment do uniformize? Our main technical contribution in this paper is to show that several cases of tractable non-uniform constraint satisfaction problems do indeed uniformize. We exhibit three non-uniform tractability results that uniformize and, thus, give rise to polynomial-time solvable cases of constraint satisfaction and conjunctive-query containment.

### Citations

11403 | Computers and Intractability: A Guide to the Theory of NP-Completeness - Garey, Johnson - 1979 |

2418 | Computational complexity - Papadimitriou - 1994 |

1597 |
Foundations of Databases
- Abiteboul, Hull, et al.
- 1995
(Show Context)
Citation Context ... m0 1 ; : : : ; m0 l ) is the leastsxed-point of the operator on A, i.e., the smallest sequence (R 1 ; : : : ; R l ) of relations on A such that (R 1 ; : : : ; R l ) = (R 1 ; : : : ; R l ) (see [AHV=-=95]-=-). We call this sequence the leastsxed-point of the system ' 1 ; : : : ; ' l and denote it by (' 1 1 ; : : : ; ' 1 l ). Usually, one is interested not in the entire sequence (' 1 1 ; : : : ; ' 1 l ), ... |

1000 | Temporal constraint networks - Dechter, Meiri, et al. - 1991 |

913 | Parameterized Complexity
- Downey, Fellows
- 1999
(Show Context)
Citation Context ...a tree decomposition of the Gaifman graph of A. Consider a tuple (a 1 ; : : : ; a n ) in a relation R of A. The elements fa 1 ; : : : ; a k g form a clique in the Gaifman graph of A. By Lemma 6.49 of =-=[DF99-=-], there is a node x of T such that fa 1 ; : : : ; a k g is contained in the label of x. It follows that T is also a tree decomposition of A. For every k 1, let A(k) be the class of allsnite relation... |

865 | Foundations of Constraint Satisfaction - Tsang - 1993 |

472 |
Optimal implementation of conjunctive queries in relational databases
- Chandra, Merlin
- 1977
(Show Context)
Citation Context ...lem as conjunctive query evaluation; moreover, conjunctive-query containment can be used as a tool in query optimization, since query equivalence is reducible to query containment. Chandra and Merlin =-=[CM77]-=- studied the computational complexity of conjunctive query containment and showed that it is an NP-complete problem. In recent years, there has been renewed interest in the study of conjunctive query ... |

446 | Information Integration using Logical Views
- Ullman
- 1997
(Show Context)
Citation Context ...d views [LMSS95, RSU95]. The latter has emerged as a central problem in integrating information from heterogeneous sources, an area that lately has been the focus of concentrated research eorts (see [=-=Ull97]-=- for survey). Since conjunctive-query containment is intractable in its full generality, researchers have embarked on a search for tractable cases. These are obtained by imposing syntactic or structur... |

438 |
Networks of constraints: Fundamental properties and applications to picture processing
- Montanari
- 1974
(Show Context)
Citation Context ... Chekuri and Rajaraman's results extend the earlier work of Yannakakis [Yan81] and Qian [Qia96] on query evaluation and containment for acyclic queries. Starting with the pioneering work of Montanari =-=[Mon74-=-] researchers in articial intelligence have investigated a class of combinatorial problems that became known as constraintsatisfaction problems (CSP). The input to such a problem consists of a set of ... |

427 | Answering Queries Using Views - Mendelzon, Srivastava |

427 |
The complexity of satisfiability problems
- Schaefer
- 1978
(Show Context)
Citation Context ...blems in an attempt to determine for which structures B the associated CSP(B) problem is tractable and for which it is intractable. The first remarkable success on this front was obtained by Schaefer =-=[Sch78]-=-, who pinpointed the computational complexity of Boolean CSP(B) problems, in which the structure B is Boolean (i.e., has the set f0; 1g as its universe). Schaefer established a dichotomy theorem for B... |

395 | Algorithms for Constraint Satisfaction Problems: A Survey - Kumar - 1992 |

381 |
Three partition refinement algorithms
- Paige, Tarjan
- 1987
(Show Context)
Citation Context ... in polynomial time. In the full paper, we will show that checking whether a relationsR is Horn or dual Horn can be done in time O(n log n), using the congruence-closure algorithm of Paige and Tarjan =-=[PT87]-=-. We say that a relation R is a trivial Schaefer relation if it is covered by the first two cases of Schaefer's classification, and we say that R is a nontrivial Schaefer relation if it is covered by ... |

322 | Linear-time algorithms for testing the satisfiability of propositional Horn formulae - Dowling, Gallier - 1984 |

288 |
Elements of the Theory of Computation
- Papadimitriou
- 1981
(Show Context)
Citation Context ...). (This implementation is inspired by the linear-time algorithms for Horn satisability [BB79, DG84].) Consider now the bijunctive case. A linear-time algorithm for 2-CNF formulas proceeds in phases [=-=LP97]-=-. In each phase, we choose an unassigned variable u and assign an arbitrary 10 truth value to it. We then use the binary clauses in the formula to propagate the assignment. If x is assigned 1 and we h... |

269 | The computational structure of monotone monadic SNP and constraint satisfaction: A study through Datalog and group theory
- Feder, Vardi
- 1998
(Show Context)
Citation Context ...l and Nesetril's dichotomy theorem generalizes the results concerning the computational complexity of the kColorability problem for each k 2. Motivated by these dichotomy results, Feder and Vardi [FV=-=99]-=- raised the question: is every CSP(B) problem either solvable in polynomial time or NP-complete? Although they did not settle this question, Feder and Vardi [FV99] were able to isolate two conditions ... |

262 |
Tree clustering for constraint networks
- Dechter, Pearl
- 1989
(Show Context)
Citation Context ...graphs of bounded-tree width (this is known as Courcelle's Theorem, see 23 [DF99]). In fact, it has already been shown in [Fre90] that these non-uniform tractability results holds uniformly (see also =-=[DP89]-=-). Here we provide a new proof of uniform tractability via a connection withsrst-order logic with a bounded number of distinct variables. Let A and B be twosnite relational structure. From Theorem 2.1... |

205 | Answering queries using templates with binding patterns - Rajaraman, Sagiv, et al. - 1995 |

200 |
A linear-time algorithm for finding tree-decompositions of small treewidth
- Bodlaender
- 1993
(Show Context)
Citation Context ... : : : ; a k g is contained in the label of x. It follows that T is also a tree decomposition of A. For every ks1, let A(k) be the class of all finite relational structures of treewidth k. Bodlaender =-=[Bod93]-=- showed that, for every ks1, there is a polynomial-time algorithm that tests whether a given graph is of treewidth k. It follows that, for every ks1, there is a polynomial-time algorithm that tests wh... |

174 |
Elementary Induction on Abstract Structures
- Moschovakis
- 1974
(Show Context)
Citation Context ... Chekuri and Rajaraman's results extend the earlier work of Yannakakis [Yan81] and Qian [Qia96] on query evaluation and containment for acyclic queries. Starting with the pioneering work of Montanari =-=[Mos74]-=-, researchers in artificial intelligence have investigated a class of combinatorial problems that became known as constraintsatisfaction problems (CSP). The input to such a problem consists of a set o... |

142 |
On the complexity of H-coloring
- Hell, Neˇsetˇril
- 1990
(Show Context)
Citation Context ...oherent explanation for the computational complexity of Horn Satis ability, 2-Satisability, One-in-Three Satisability, and other such Boolean satisability problems. After this, Hell and Nesetril [HN90=-=]-=- established a dichotomy theorem for CSP(B) 3 problems in which B is an undirected graph: if B is 2-colorable, then CSP(B) is solvable in polynomial time; otherwise, CSP(B) is NP-complete. Observe tha... |

141 | Query folding
- Qian
- 1996
(Show Context)
Citation Context ...a graph (see [vL90, Bod93]). It should be noted that queries of width 1 are precisely the acyclic queries; thus, Chekuri and Rajaraman's results extend the earlier work of Yannakakis [Yan81] and Qian =-=[Qia96-=-] on query evaluation and containment for acyclic queries. Starting with the pioneering work of Montanari [Mon74] researchers in articial intelligence have investigated a class of combinatorial proble... |

133 | Hypertree decompositions and tractable queries
- Gottlob, Leone, et al.
(Show Context)
Citation Context ...ph is also what they called query decomposition. (Note, however, that the property of having treewidth k can be tested in linear time [Bod93], while the property of having querywidth 4 is NP-complete =-=[GLS99-=-].) Thus, the polynomial tractability of CSP(A(k); F) follows also from the results in [CR98]. Gottlob, Leone, and Scarcello [GLS99] dene yet another notion of width, called hypertree width. They show... |

132 | On local and nonlocal properties - Gaifman - 1982 |

118 | Computational problems related to the design of normal form relational schemes - BEERI, BERNSTEIN - 1979 |

106 | On the algebraic structure of combinatorial problems
- Jeavons
- 1998
(Show Context)
Citation Context ...f constraint satisfaction, a perusal of the literature reveals that all constraintsatisfaction problems studied can be viewed as special cases of the above homomorphism problem [FV93, FV99] (see also =-=[Jea97]-=-). It should be noted that several researchers, including [Bib88, Dec90, GJC94, PJ97], have observed that there are tight connections between constraint-satisfaction problems and certain problems in r... |

98 | Conjunctive query containment revisited
- Chekuri, Rajaraman
- 1997
(Show Context)
Citation Context ... (LOGCFL is the class of decision problems that are logspace-reducible to a context-free language). We note that another way to dene the treewidth of a structure A is in terms of its incidence graph [=-=CR97]-=-. The incidence graph of A is a bipartite graph that has all the tuples in relations of A as nodes in one part, the elements of the universe of A as nodes in the other part, and there is an edge from ... |

92 | On the conversion between Non-Binary and Binary constraint satisfaction problems
- Bacchus, Beek
- 1998
(Show Context)
Citation Context ....e., structures with binary relations only). We refer to the binary encoding of a structure A by binary(A). The reduction from A to binary(A) used here is the dual-graph representation of [DP89]. See =-=[BB98]-=- for experimental results concerning this reduction, as well as for another reduction, called the hidden-variable translation. 26 The vocabulary of binary(A) contains a binary relation symbol E P;Q;i;... |

85 | Decomposing constraint satisfaction problems using database techniques
- Gyssens, Jeavons, et al.
- 1994
(Show Context)
Citation Context ...8, Dec90, GJC94, PJ97], have observed that there are tight connections between constraint-satisfaction problems and certain problems in relational databases. In particular, Gyssens, Jeavons and Cohen =-=[GJC94]-=- pointed out that the set of all solutions to a constraint satisfaction problem coincides with the join of certain relations extracted from the given constraint-satisfaction problem. Thus, solving con... |

80 |
Complexity of K-tree structured constraint satisfaction problems
- Freuder
- 1990
(Show Context)
Citation Context ...hs denable in monadic second-order logic is decidable in polynomial time for graphs of bounded-tree width (this is known as Courcelle's Theorem, see 23 [DF99]). In fact, it has already been shown in [=-=Fre90]-=- that these non-uniform tractability results holds uniformly (see also [DP89]). Here we provide a new proof of uniform tractability via a connection withsrst-order logic with a bounded number of disti... |

78 | Structure identification in relational data
- Dechter, Pearl
- 1992
(Show Context)
Citation Context ...). In addition, Schaefer showed that a Boolean relation R is affine if and only if the following condition holds: if t 1 ; t 2 ; t 3 2 R, then (t 1 \Phi t 2 \Phi t 3 ) 2 R. Finally, Dechter and Pearl =-=[DP92]-=- showed that a Boolean relation R is Horn (resp., dual Horn) if and only if the following condition holds: if t 1 ; t 2 2 R, then t 1st 2 2 R (resp., t 1st 2 2 R). 2 Clearly, each of these conditions ... |

77 | The complexity of acyclic conjunctive queries
- Gottlob, Leone, et al.
(Show Context)
Citation Context .... As shown in [Var95], 9FO k+1 has polynomial-time combined complexity, which implies that CSP(A(k); F) is solvable in polynomial time. 25 A precise complexity analysis of CSP(A(k); F) is provided in =-=[GLS98-=-], where it is shown that the problem is LOGFCL-complete (LOGCFL is the class of decision problems that are logspace-reducible to a context-free language). We note that another way to dene the treewid... |

77 |
On the complexity of bounded-variable queries
- Vardi
- 1995
(Show Context)
Citation Context ... can be found in time polynomial in the size of A. Thus, in this case, checking the existence of a homomorphism h : A ! B reduces to the evaluation of an 9FO k+1 query on the structure B. As shown in =-=[Var95]-=-, 9FO k+1 has polynomial-time combined complexity, which implies that CSP(A(k); F) is solvable in polynomial time. 25 A precise complexity analysis of CSP(A(k); F) is provided in [GLS98], where it is ... |

73 |
clause queries and generalizations
- Horn
- 1985
(Show Context)
Citation Context ...d-points of systems of positive FO k -formulas. Similarly, 9LFP k is the fragment of LFP obtained by taking the componets of leastsxed-points of systems of positive 9FO k -formulas. Chandra and Harel =-=[CH85-=-] showed that Datalog has the same expressive power as the existential fragment of LFP. More precisely, a query is expressible in k-Datalog if and only if it is 9LFP k -denable. In fact, every k-Datal... |

61 | Upper bounds to the clique-width of graphs
- Courcelle, Olariu
(Show Context)
Citation Context ...at its treewidth is at most k." Is OPT decidable in polynomial time? A class that is more general than the class of bounded treewidth graphs is the class of bounded cliquewidth graphs. It is show=-=n in [CO98]-=- than if a graph has treewidth k, then its cliquewidth is bounded from above by 2 k+1 + 1. Thus, a class of graphs that has bounded treewidth also has bounded cliquewidth. Courcelle, Makowsky, and Rot... |

56 | On the expressive power of Datalog: tools and a case study
- Kolaitis, Vardi
- 1995
(Show Context)
Citation Context ...actability of many nonuniformsCSP(B) problems can be globally explained by the fact that the complement of each of these problems is expressible in Datalog. Using pebble-game techniques introduced in =-=[KV95-=-], we show here that such non-uniform tractability results uniformize, as long as 4 Datalog programs with a bounded number of distinct variables are considered. Specically, we establish that, for ever... |

53 |
Monotone monadic SNP and constraint satisfaction
- Feder, Vardi
- 1993
(Show Context)
Citation Context ... cases of CSP(A; B) that are obtained by imposing restrictions on the class A. For this, we consider the concept of treewidth of a relational structure; this concept was introduced by Feder and Vardi =-=[FV93]-=- and generalizes the concept of treewidth of a graph (see [vL90, Bod93]). A tree decomposition of asnite relational structure A is a labeled tree T such that the following conditions hold: 1. every no... |

51 | Constraint Satisfaction Problems: An Overview - Meseguer - 1989 |

51 | Graph algorithms - Leeuwen - 1992 |

48 | The decision problem for some classes of sentences without quantifiers - McKinsey - 1943 |

45 | A survey of tractable constraint satisfaction problems
- Pearson, Jeavons
- 1997
(Show Context)
Citation Context ...s has pursued tractable cases of constraint satisfaction CSP(A; B) by investigating the class of functions under which the relations in the structures in B are closed [JC95, JCG95, JCG96, Jea97] (see =-=[PJ97]-=- for a survey). In [FV93, FV99], a preliminary investigation has been carried out on the connection between expressibility of CSP(B) problems in Datalog and closure of the relations in B under certain... |

43 | Infinitary logic and 01 laws
- Kolaitis, Vardi
- 1992
(Show Context)
Citation Context ...LFP is subsumed by thesnite-variable innitary logic L ! 1! = S k L k 1! , where L k 1! is the innitary logic with arbitrary disjunctions and conjunctions, but with at most k distinct variables (see [K=-=V92-=-]). In the present paper, we are interested in fragments of L k 1! and L ! 1! that are suitable for the study of Datalog. For every k 1, let 9L k 1! be the existential positive fragment of L ! 1! wit... |

36 | Linear-time algorithms for testing the satis of propositional horn formulae - Dowling, Gallier - 1984 |

34 |
The complexity of satis problems
- Schaefer
- 1978
(Show Context)
Citation Context ...roblems in an attempt to determine for which structures B the associated CSP(B) problem is tractable and for which it is intractable. Thesrst remarkable success on this front was obtained by Schaefer =-=[Sch78]-=-, who pinpointed the computational complexity of Boolean CSP(B) problems, in which the structure B is Boolean (i.e., has the set f0; 1g as its universe). Schaefer established a dichotomy theorem for B... |

31 | A game-theoretic approach to constraint satisfaction - Kolaitis, Vardi |

29 | A unifying framework for tractable constraints - Jeavons, Cohen, et al. - 1995 |

28 | The complexity of constraint satisfaction revisited - Mackworth, Freuder - 1993 |

27 |
Node-deletion problems on bipartite graphs
- Yannakakis
- 1981
(Show Context)
Citation Context ... of treewidth of a graph (see [vL90, Bod93]). It should be noted that queries of width 1 are precisely the acyclic queries; thus, Chekuri and Rajaraman's results extend the earlier work of Yannakakis =-=[Yan81-=-] and Qian [Qia96] on query evaluation and containment for acyclic queries. Starting with the pioneering work of Montanari [Mon74] researchers in articial intelligence have investigated a class of com... |

23 | A test for tractability - Jeavons, Cohen, et al. - 1996 |

22 | Constraint satisfaction from a deductive viewpoint - Bibel - 1988 |