C. Li and E. Chang. Query planning with limited source capabilities (extended version). Technical report, Computer Science Dept., Stanford Univ., 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....systems where the global schema is expressed in terms of an extended Entity Relationship model. With regard to the specification of the mapping between the global schema and the sources, two basic approaches have been used to specify the mapping between the sources and the global schema [13, 15, 16]. The first approach, called global as view (also global schema centric, or simply global centric) requires that the global schema is expressed in terms of the data sources. The second approach, An extended version of this paper was published in the Proc. of the 20th Int. Conf. on Conceptual ....

Chen Li and Edward Chang. Query planning with limited source capabilities. In Proc. of ICDE 2000, pages 401--412, 2000.

....Thus, our third contribution is the implementation and experimental validation of the mediator and its heuristics. There are several areas of research that are relevant to our work. There has been much work in capability based rewriting of mediator queries with limited query capability of sources [14, 15, 20, 27, 28, 29, 34, 47, 48, 49, 51]. While we use the results of this research, this paper does not directly make a contribution in this area. There has also been some research on estimating the costs for accessing heterogeneous sources [1, 11, 31, 32] There has also been some research on considering both capability and costs in ....

....there are some WebSources and WSIs that can provide an answer. b) For each mediator subgoal in the query, it identifies all the relevant WSIs. The problem solved by the CBR Tool is the AcceptedQuery problem [49] and several heuristic solutions to this problem and related problems are available [14, 15, 20, 27, 28, 29, 34, 47, 48, 49, 51]. While our optimizer uses capability based rewriting, this is not the focus of this paper, and we provide a brief overview of the CBR Tool in this section. Consider the following query expressed in an SQL like syntax: From Paper, CoAuthor, Editor, Reviewer Where 1stAuthor= Franklin and ....

C. Li and E. Chang. Query planning with limited source capabilities. Proceedings of ICDE, 2000.

....Here, we concentrate on the following issues: 1. Specifying the mapping between the global schema and the sources, 2. Processing queries expressed on the global schema. With regard to issue (1) two basic approaches have been used to specify the mapping between the sources and the global schema [9 11]. The rst approach, called global as view (or global centric) requires that the global schema is expressed in terms of the data sources. More precisely, to every element of the global schema, a view over the data sources is associated, so that its meaning is speci ed in terms of the data ....

Chen Li and Edward Chang. Query planning with limited source capabilities. In Proc. of the 16th IEEE Int. Conf. on Data Engineering (ICDE

....issues [10] One of the most important aspect is the speci cation of the mapping between the global schema and the sources, and the use of such a speci cation for carrying out query processing. Two basic approaches have been used to specify the mapping between the sources and the global schema [11, 12, 13]. The rst approach, called global as view (or simply GAV) requires that the global schema is expressed in terms of the data sources. More precisely, to every element of the global schema, a view over the data sources is associated, so that its meaning is speci ed in terms of the data residing at ....

Chen Li and Edward Chang. Query planning with limited source capabilities. In Proc. of the 16th IEEE Int. Conf. on Data Engineering (ICDE

....Here, we concentrate on two basic issues: 1. specifying the mapping between the global schema and the sources, 2. processing queries expressed on the global schema. With regard to issue (1) two basic approaches have been used to specify the mapping between the sources and the global schema [19, 22, 23]. The rst approach, called globalcentric, requires that the global schema is expressed in terms of the data sources. More precisely, to every element of the global schema, a view over the data sources is associated, so that its meaning is speci ed in terms of the data residing at the sources. In ....

Chen Li and Edward Chang. Query planning with limited source capabilities. In Proc. of the 16th IEEE Int. Conf. on Data Engineering (ICDE

....idea, and investigate the problem of query answering in data integration systems where the global schema is expressed in terms of an extended Entity Relationship Model. With regard to issue (ii) two basic approaches have been used to specify the mapping between the sources and the global schema [15, 17, 18]. The first approach, called global as view (also global schema centric, or simply global centric) requires that the global schema is expressed in terms of the data sources. More precisely, to every concept of the global schema, a view over the data sources is associated, so that its meaning is ....

C. Li and E. Chang. Query planning with limited source capabilities. In Proc. of ICDE

No context found.

C. Li and E. Chang. Query planning with limited source capabilities (extended version). Technical report, Computer Science Dept., Stanford Univ., 1999.

No context found.

C. Li and E. Chang. Query planning with limited source capabilities. International Conference on Data Engineering (ICDE), 2000.

....to compute mediator restrictions given source restrictions. These studies did not consider the possibility that removing subgoals may make an infeasible query feasible. Thus, they regard the query Q 2 in Example 1.2 as an unsolvable query, thus miss the chances of computing its complete answer. LC00, LC01a] studied how to compute the maximal answer to a conjunctive query with binding restrictions by borrowing bindings from relations not in the query. The paper focused on how to trim irrelevant relations that do not help in obtaining bindings. However, the computed answer may not be the ....

....develop a decision tree that guides the planning process to compute the complete answer to a CQ. We discuss two planning strategies that can be taken while traversing the tree. We also discuss how to optimize a CQ to compute its complete answer. 7. 1 Dynamic Computability of Complete Answers In [LC00, LC01a] we show that we can compute an answer to a query by borrowing bindings from relations not in the query, even though the borrowed bindings do not guarantee to retrieve tuples from relations. For instance, suppose a Web search form requires a book title to return book information. We can go ....

[Article contains additional citation context not shown here]

Chen Li and Edward Chang. Query planning with limited source capabilities. In ICDE, pages 401--412, 2000.

....with limited access patterns, how to test whether the maximal answer to the first query is contained in the maximal answer to the second one Clearly the solution to this problem can be used to answer queries efficiently. Given a conjunctive query on relations with limited access patterns, [8, 17] show how to construct a recursive datalog program [24] to compute the maximal answer to the query. That is, we can retrieve tuples from relations by retrieving as many bindings from the relations and the query as possible, then use the obtained tuples to answer the query. For instance, consider ....

....of Phi a . a) If I a is empty (arc 5) then we know that the complete answer is empty. b) If I a is not empty (arc 6) let I P a be the projection of I a onto the distinguished variables. We use all the bindings from the query and the relations to retrieve as many tuples as possible. See [8, 17] for the details. Let Phi na denote all the nonanswerable subgoals. i. If for every tuple t P 2 I P a , there is a tuple t a 2 I a , such that the projection of t a onto the distinguished variables is t P , and t a can join with some tuples for all the subgoals Phi na (tuple t P is ....

[Article contains additional citation context not shown here]

C. Li and E. Chang. Query planning with limited source capabilities. In ICDE, pages 401--412, 2000.

....2. How to compute the capabilities of mediators on relations with limited capabilities [12] We consider other complicated relation capabilities besides binding restrictions. 3. How to do query reformulation to compute as many answers to a query as possible in the presence of binding restrictions [5]. We show that a query can be answered by borrowing bindings from sources not mentioned in the query. We also develop an algorithm for finding all the relations that need to accessed to answer a query. 4. How to test whether by query reformulation, it is possible to compute the complete to a ....

C. Li and E. Chang. Query planning with limited source capabilities. International Conference on Data Engineering (ICDE), 2000.

....an EMS of V . As shown in Example 4.1, a view set can have multiple EMS s. In some scenarios we need to answer queries using views in the presence of constraints (e.g. functional dependencies [Cod70] multivalued dependencies [Fag77, Del78] and binding limitations on views [RSU95, DL97, Ull89, LC00, LYV 98, YLUGM99] We can generalize our results by requiring that the algorithm Check F take constraints and binding limitations into consideration. RSU95, Gry99] give algorithms for answering conjunctive queries using conjunctive views in these scenarios; Dus97] provides answers for ....

Chen Li and Edward Chang. Query planning with limited source capabilities. International Conference on Data Engineering (ICDE), pages 401-412, 2000.

....a query means a legal plan for the query. Definition 2.3 (exhaustive plan) Let Q be a query on a set of relations R. The exhaustive plan for Q is a plan that accesses all the possible relations R to obtain bindings, and retrieves as many tuples from the relations as possible to answer the query. [7, 20] showed how to implement an exhaustive plan by translating the query Q and the relations R with their binding patterns into a datalog program, and evaluating the program on the relations R. 2 Definition 2.4 (observable tuples) Given a database of relations R, a tuple t of a relation r i 2 R is ....

....32] studied query optimization in the presence of binding restrictions. 31] considered how to compute mediator restrictions given source restrictions. These four studies did not minimize a CQ before checking its feasibility. Thus, they regard the query Q 2 in Example 1.2 as an unsolvable query. [20] studied how to compute the maximal answer to a CQ in the presence of binding restrictions by borrowing bindings from relations not in the query. However, the computed answer may not be the complete answer. As we will see in Section 4, we can sometimes use the approach in that paper to compute the ....

C. Li and E. Chang. Query planning with limited source capabilities. International Conference on Data Engineering (ICDE), pages 401--412, 2000.

....compute mediator restrictions given source restrictions. These four studies did not consider the possibility that removing subgoals may make an infeasible query feasible. Thus, they regard the query Q 2 in Example 1.2 as an unsolvable query, thus miss the chances of computing its complete answer. LC00] studied how to compute the maximal answer to a conjunctive query with binding restrictions by borrowing bindings from relations not in the query. The paper focused on how to trim irrelevant relations that do not help in obtaining bindings. However, the computed answer may not be the complete ....

....is the complete answer. 2 Example 2.3 shows that relations not mentioned in a query can be accessed to contribute bindings to compute the query s results. An exhaustive plan accesses all possible relations to obtain as many bindings as possible, and answer the query using the retrieved tuples. LC00] showed how to implement an exhaustive plan by translating restrictions and a CQ into a Datalog program, and evaluating the program on the relations. Datalog is used in the planning process since the process can be recursive (i.e. we may access the relations repeatedly to obtain more bindings) ....

[Article contains additional citation context not shown here]

Chen Li and Edward Chang. Query planning with limited source capabilities. International Conference on Data Engineering (ICDE), 2000.

....25] studied query optimization in the presence of binding restrictions. 24] considered how to compute mediator restrictions given source restrictions. These four studies did not minimize a CQ before checking its feasibility. Thus, they regard the query Q 2 in Example 1.2 as an unsolvable query. [15] studied how to compute the maximal answer to a CQ in the presence of binding restrictions by borrowing bindings from relations not in the query. However, the computed answer may not be the complete answer. As we will see in Section 4, we can sometimes use the approach in that paper to compute the ....

....about the possibility of computing the complete answer, and it traverses one more level by doing the corresponding operations. 2 An exhaustive plan for a query accesses all possible relations to obtain bindings, and retrieves as many tuples from the relations as possible to answer the query. [15] showed how to implement an exhaustive plan by translating source restrictions and a CQ into a Datalog program, and evaluating the program on the relations. 6 use a linear plan following complete answer the order to compute the variables bound in a use a linear plan to compute I a all ....

C. Li and E. Chang. Query planning with limited source capabilities. International Conference on Data Engineering (ICDE), 2000.

....Containment in the Presence of Binding Restrictions (Technical Report) Chen Li Department of Computer Science Stanford University chenli db.stanford.edu December 22, 1999 Abstract In information integration systems, sources have diverse and limited query capabilities. In a recent paper [LC00] we showed that sources not mentioned in a query can contribute to the query result by providing useful bindings. We studied connection queries, where each connection query is a natural join of distinct source views with the necessary selection and projection. Some optimization problems are ....

....join of distinct source views with the necessary selection and projection. Some optimization problems are left open, including whether the answer computed by one connection is contained in that computed by another connection. In this paper we study this connection containment problem. Since [LC00] often produces a recursive Datalog program to answer a connection query optimally, containment seems undecidable. However, because the Datalog programs of [LC00] have a special form, their containment can be reduced to containment of monadic programs, which is known to be decidable. Further, ....

[Article contains additional citation context not shown here]

Chen Li and Edward Chang. Query planning with limited source capabilities. International Conference on Data Engineering (ICDE), 2000. 16

....sequence: v 1 , v 3 . The following theorem shows that an independent connection does not require bindings from views outside the connection. Due to space limitations, we have not provided all the proofs of the lemmas and theorems in this paper. They are available in the full version of the paper [15]. Theorem 4.1 If connection T is independent, then for any source relations, we can compute the complete answer for T by using only the source views in T . 2 Theorem 4.2 For a nonindependent connection T , there exists an instance of the source relations, such that some tuples in the complete ....

...., v 3 , v 4 . 5.4. The algorithm FIND REL Now we show how to find all the relevant views of a connection by giving the following theorem: Theorem 5.1 If K is a kernel of a queryable connection T , then b closure(K) # T are all the relevant source views of connection T . 2 Proof: Refer to [15] for the detailed proof. The essential idea is that we need to prove, for a kernel K of a queryable connection T : i) All the source views in V b closure(K)# T are irrelevant to T ; ii) Every source view in T is relevant to T ; iii) Every source view in b closure(K) is relevant to T . We ....

C. Li and E. Chang. Query planning with limited source capabilities (extended version). Technical report, Computer Science Dept., Stanford Univ., 1999.

No context found.

Chen Li and Edward Chang. Query planning with limited source capabilities. In Proc. of the 16th IEEE Int. Conf. on Data Engineering (ICDE 2000.

No context found.

Chen Li and Edward Chang. Query planning with limited source capabilities. In Proc. of the 16th IEEE Int. Conf. on Data Engineering (ICDE 2000.

No context found.

Chen Li and Edward Chang. Query planning with limited source capabilities. In Proc. of the 16th IEEE Int. Conf. on Data Engineering (ICDE 2000), pages 401--412, 2000.

No context found.

Li, C., Chang, E.: Query planning with limited source capabilities. In: Proc. of ICDE 2000. (2000) 401--412

No context found.

C. Li and E. Chang. Query planning with limited source capabilities. In Proc. of ICDE 2000, pages 401--412, 2000.

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