## The Complexity of Querying Indefinite Data about Linearly Ordered Domains (1992)

Venue: | In The Proceedings of the Eleventh ACM SIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems |

Citations: | 40 - 2 self |

### BibTeX

@INPROCEEDINGS{Meyden92thecomplexity,

author = {Ron Van Der Meyden},

title = {The Complexity of Querying Indefinite Data about Linearly Ordered Domains},

booktitle = {In The Proceedings of the Eleventh ACM SIGACT-SIGMOD-SIGART Symposium on Principles of Database Systems},

year = {1992},

pages = {331--345}

}

### Years of Citing Articles

### OpenURL

### Abstract

In applications dealing with ordered domains, the available data is frequently indefinite. While the domain is actually linearly ordered, only some of the order relations holding between points in the data are known. Thus, the data provides only a partial order, and query answering involves determining what holds under all the compatible linear orders. In this paper we study the complexity of evaluating queries in logical databases containing such indefinite information. We show that in this context queries are intractable even under the data complexity measure, but identify a number of PTIME sub-problems. Data complexity in the case of monadic predicates is one of these PTIME cases, but for disjunctive queries the proof is non-constructive, using well-quasi-order techniques. We also show that the query problem we study is equivalent to the problem of containment of conjunctive relational database queries containing inequalities. One of our results implies that the latter is \Pi p 2 ...

### Citations

10921 |
Computers and Intractability: A Guide to the Theory of NP-Completeness
- Garey, Johnson
(Show Context)
Citation Context ... Theorem 3.2: There exists a conjunctive f!g-query containing only binary predicates with co-NP hard data complexity on f!g-databases. Proof: The proof is by reduction from monotone 3--satisfiability =-=[10]-=-. We show that there exists a query \Phi and a polynomial time reduction from sets of monotone 3--clauses S to f!g-databases D such that D j= \Phi if and only if S is unsatisfiable. Given object const... |

2284 | Maintaining knowledge about temporal intervals
- Allen
- 1983
(Show Context)
Citation Context ...oural psychology, biology, scheduling problems in operations research, and combinatorics. Indefinite order information also arises naturally in a variety of contexts in artificial intelligence. Allen =-=[2]-=- has pointed out that in natural language most temporal reports describe relations between intervals, rather than give absolute times. He proposes an algebra based on thirteen primitive temporal relat... |

1132 |
Algorithmic graph theory and perfect graphs
- Golumbic
- 1980
(Show Context)
Citation Context ...d to have been in use for some historical interval. Absolute data for these intervals is rarely available, but coincidence of two artifacts in a grave indicates that their intervals overlap. Golumbic =-=[12]-=- describes this and many other examples of indefinite order data in various domains, including behavioural psychology, biology, scheduling problems in operations research, and combinatorics. Indefinit... |

530 |
The complexity of relational query languages
- Vardi
- 1982
(Show Context)
Citation Context ... databases, aside from the related work of Klug already mentioned. Our contribution in this paper is to provide such an analysis. We consider the following measures of complexity, introduced by Vardi =-=[31]-=-. Combined complexity is the complexity of the set fhD; \PhiijD j= \Phig. This measures complexity of query answering as a function of both the size of the query 1 Kanellakis et al. [15] have previous... |

451 |
Model Theory
- Chang, Keisler
- 1990
(Show Context)
Citation Context ...r all M 2 ModO (D): For the restricted form of database and query we are considering, these consequence relations are closely related, as we now show. First, we need a standard model theoretic notion =-=[6]-=-. A homomorphism from a model M to a model M 0 is a mapping h from the domain of M to the domain of M 0 such that ffl If a is of sort object (order) then h(a) is of sort object (order). ffl If the int... |

448 |
Optimal implementation of conjunctive queries in relational data bases
- Chandra, Merlin
- 1977
(Show Context)
Citation Context ...n a query Q2 if in every relational database (containing only definite information) the answer (set of tuples) to Q1 is a subset of the answer to Q2. Klug noted that the classical homomorphism theory =-=[5]-=- for containment of conjunctive queries, which shows that this problem is NP complete, does not extend to queries containing inequalities (although it does work for a subset of these queries, the semi... |

372 | P.: “Constraint propagation algorithms for temporal reasoning: a revised report
- Vilain, Kautz, et al.
- 1989
(Show Context)
Citation Context ...sidered in the context of applications to constraint logic programming [21, 29]. The complexity of reasoning about order relations has also received some attention in the AI literature. Vilain et al. =-=[33]-=- study the complexity of inferring relationships between intervals in Allen's interval algebra, that is, they study the complexity of determining whether an interval relationship I 1 RI 2 follows, whe... |

336 | Constraint query languages
- Kanellakis, Kuper, et al.
- 1995
(Show Context)
Citation Context ...uced by Vardi [31]. Combined complexity is the complexity of the set fhD; \PhiijD j= \Phig. This measures complexity of query answering as a function of both the size of the query 1 Kanellakis et al. =-=[15]-=- have previously shown that containment of conjunctive queries using quadratic equation constraints is \Pi p 2 -complete. They also consider data complexity for a notion of (definite) constraint datab... |

216 |
Towards a logical reconstruction of relational database theory
- Reiter
- 1984
(Show Context)
Citation Context ...e database as a first order theory. We reserve the word `points' to refer to elements of the order sort; elements of the object sort will be called `objects'. We do not make a unique names assumption =-=[26]-=- in this paper: distinct constants may refer to the same point. For order constants this is because we explicitly wish to allow distinct order constants to refer to the same point; for object constant... |

206 |
Principles of Database and Knowledge Base Systems
- Ullman
- 1989
(Show Context)
Citation Context ...point based language with relations `!', `' and `6='. The problem of deriving point relationships, that is, the problem of determining if uRv follows for R 2 f!; ; 6=g, has polynomial time complexity =-=[30, 3]-=-. Golumbic and Shamir [11] present a finer grained analysis of the complexity of inferring interval relations in the interval algebra, showing the effect on complexity of various restrictions on the s... |

137 |
On conjunctive queries containing inequalities
- Klug
- 1988
(Show Context)
Citation Context ...the problem we consider in terms of indefinite information, but it is closely related to optimization problems for relational database queries containing inequalities, which have been studied by Klug =-=[17]-=-. A query Q1 is said to be contained in a query Q2 if in every relational database (containing only definite information) the answer (set of tuples) to Q1 is a subset of the answer to Q2. Klug noted t... |

119 | A propositional modal logic of time intervals
- Halpern, Shoham
- 1991
(Show Context)
Citation Context ... disjunctive normal form representation is possible. 8 Conclusion Our results in this paper conform to a pattern noted elsewhere: reasoning about intervals is more complex than reasoning about points =-=[13, 33]-=-. We have seen 6 Closely related results appear in [1, 32] for n-ary predicates and less restricted query forms. that binary predicates, which in combination with the order relations permit the repres... |

115 | On the representation and querying of sets of possible worlds
- Abiteboul, Kanellakis, et al.
- 1991
(Show Context)
Citation Context ...proper predicates as well as the order predicates `!' and `'. While much is already known about the complexity of querying indefinite databases containing null values subject to `6=' constraints only =-=[32, 1]-=-, there does not appear to have been any analysis of the complexity of this more general class of queries in indefinite order databases, aside from the related work of Klug already mentioned. Our cont... |

86 | Complexity and algorithms for reasoning about time: a graph-theoretic approach
- Golumbic, Shamir
- 1993
(Show Context)
Citation Context ...ations `!', `' and `6='. The problem of deriving point relationships, that is, the problem of determining if uRv follows for R 2 f!; ; 6=g, has polynomial time complexity [30, 3]. Golumbic and Shamir =-=[11]-=- present a finer grained analysis of the complexity of inferring interval relations in the interval algebra, showing the effect on complexity of various restrictions on the set of primitive relations.... |

77 |
Exact and Approximate Reasoning about Temporal Relations
- Beek, Cohen
- 1990
(Show Context)
Citation Context ...point based language with relations `!', `' and `6='. The problem of deriving point relationships, that is, the problem of determining if uRv follows for R 2 f!; ; 6=g, has polynomial time complexity =-=[30, 3]-=-. Golumbic and Shamir [11] present a finer grained analysis of the complexity of inferring interval relations in the interval algebra, showing the effect on complexity of various restrictions on the s... |

70 |
The theory of well-quasi-ordering: a frequently discovered concept
- Kruskal
(Show Context)
Citation Context ... not exist indices i ! j such that x i �� x j . Then X is well-quasiordered if and only if there does not exist a bad infinite sequence. The theory of well-quasi-orders is well developed: see Krus=-=kal [19] for-=- an historical survey and Milner [25] for a more detailed introduction. If X is a quasi-ordered set then we may define a quasi-order on the set FP(X) of finite subsets of X by S 1 �� S 2 when for ... |

69 |
A Structure for Plans and Behaviour
- Sacerdoti
- 1977
(Show Context)
Citation Context ...al I overlaps interval J" and gives an algorithm for making (incomplete) inferences about interval relations. Another example of indefinite order data in artificial intelligence is nonlinear plan=-=ning [28]-=-. Here, rather than the solution to a planning problem being a linear sequence of actions, one constructs a partially ordered set of actions. This allows some flexibility in the order of execution of ... |

54 |
Processing conjunctive predicates and queries
- Rosenkrantz, H
- 1980
(Show Context)
Citation Context ...ts, which includes a number of studies of complexity issues. In the database field, the problem of inferring inequalities from other inequalities has been studied in connection with predicate locking =-=[27]-=-. The more general problem of inferring linear inequalities of the form ax+by+ : : :sp from other such inequalities has been considered in the context of applications to constraint logic programming [... |

49 |
der Meyden. The Complexity of Querying Indefinite Data about Linearly Ordered Domains
- van
- 1997
(Show Context)
Citation Context ... maximal class with respect to the parameters we have considered in this paper for combined complexity of monadic queries. For the proofs of Proposition 5.4 and Proposition 5.5 we refer the reader to =-=[23]-=-. 6 Data Complexity of Disjunctive Monadic Queries We have seen that the data complexity of conjunctive monadic queries is in polynomial time. The data complexity of disjunctive monadic queries also t... |

37 |
An overview of sequence comparison. In Time warps, string edits, and macromolecules: The theory and practice of sequence comparison
- Kruskal
- 1983
(Show Context)
Citation Context ...rk for a more expressive class of queries. It would be interesting to explore the relationships between this result and the dynamic programming techniques used to find optimal alignments of sequences =-=[20]-=-. Equivalent definitions are as follows. First, a set is well-quasi-ordered just in case (a) every strictly decreasing sequence is finite and (b) every set of pairwise incomparable elements is finite.... |

35 | Subsumption and indexing in constraint query languages with linear arithmetic constraints
- Srivastava
- 1993
(Show Context)
Citation Context ...]. The more general problem of inferring linear inequalities of the form ax+by+ : : :sp from other such inequalities has been considered in the context of applications to constraint logic programming =-=[21, 29]-=-. The complexity of reasoning about order relations has also received some attention in the AI literature. Vilain et al. [33] study the complexity of inferring relationships between intervals in Allen... |

35 |
Querying logical databases
- Vardi
- 1985
(Show Context)
Citation Context ...proper predicates as well as the order predicates `!' and `'. While much is already known about the complexity of querying indefinite databases containing null values subject to `6=' constraints only =-=[32, 1]-=-, there does not appear to have been any analysis of the complexity of this more general class of queries in indefinite order databases, aside from the related work of Klug already mentioned. Our cont... |

29 |
Basic wqo and bqo theory
- MILNER
(Show Context)
Citation Context ... x j . Then X is well-quasiordered if and only if there does not exist a bad infinite sequence. The theory of well-quasi-orders is well developed: see Kruskal [19] for an historical survey and Milner =-=[25] for-=- a more detailed introduction. If X is a quasi-ordered set then we may define a quasi-order on the set FP(X) of finite subsets of X by S 1 �� S 2 when for each x 2 S 1 there exists y 2 S 2 with x ... |

24 |
Dense time and temporal constraints with 6
- Koubarakis
- 1992
(Show Context)
Citation Context ... the set of order atoms entailed by a f!; ; 6=g-database (see Ullman [30] Section 14.2, van Beek and Cohen [3]), this set does not capture all the disjunctive consequences of the database. Koubarakis =-=[18]-=- has shown that it is possible to project a set of f!; ; 6=gconstraints onto a polynomial size representation which contains a number of disjunctions of inequations. However, we do not know whether a ... |

22 |
Querying Constraints
- Lassez
- 1990
(Show Context)
Citation Context ...]. The more general problem of inferring linear inequalities of the form ax+by+ : : :sp from other such inequalities has been considered in the context of applications to constraint logic programming =-=[21, 29]-=-. The complexity of reasoning about order relations has also received some attention in the AI literature. Vilain et al. [33] study the complexity of inferring relationships between intervals in Allen... |

10 |
der Meyden. Recursively indefinite databases
- van
- 1990
(Show Context)
Citation Context ... of databases capable of expressing interval data. In fact, with this constraint, it is even possible to combine the order indefiniteness with certain forms of recursive indefiniteness, introduced in =-=[24]-=-, while retaining tractability. The techniques used in this work generalize the ideas of Theorem 5.3. Acknowledgements: This paper represents research done primarily at Rutgers University and complete... |

6 |
der Meyden. Reasoning about indefinite actions
- McCarty, van
- 1992
(Show Context)
Citation Context ... the representation of certain forms of interval data, lead to increased complexity. Elsewhere, we show that the picture on interval data is actually not so bleak as it appears from the present paper =-=[23, 22]-=-. By replacing the bounded width constraint by a slightly more restrictive condition, called bounded concurrency, we obtain a tractable class of databases capable of expressing interval data. In fact,... |

3 |
Nonconstructive advances in polynomial time complexity
- Fellows, Langston
- 1987
(Show Context)
Citation Context ...that combined complexity is co-NP hard indicates that the constants of proportionality may be very large. 4 A number of other examples of non-constructive proofs that a set is in PTIME are known, see =-=[8, 9]-=-. One special case in which we do know how to compute a basis is when the query \Phi is conjunctive. Lemma 4.1 and Lemma 4.2 together show that a conjunctive monadic query \Phi is entailed by a databa... |

3 |
Nonconstructive tools for proving polynomial time decidability
- Fellows, Langston
- 1988
(Show Context)
Citation Context ...that combined complexity is co-NP hard indicates that the constants of proportionality may be very large. 4 A number of other examples of non-constructive proofs that a set is in PTIME are known, see =-=[8, 9]-=-. One special case in which we do know how to compute a basis is when the query \Phi is conjunctive. Lemma 4.1 and Lemma 4.2 together show that a conjunctive monadic query \Phi is entailed by a databa... |

2 |
Some methods and problems in statistical archeology
- Kendall
- 1969
(Show Context)
Citation Context ...nst either suspect. 2 Many applications give rise to indefinite data about linear order. As in the example, the linearly ordered domain is often a time line. In the problem of seriation in archeology =-=[16]-=- each type of artifact is assumed to have been in use for some historical interval. Absolute data for these intervals is rarely available, but coincidence of two artifacts in a grave indicates that th... |