W. Kent, R. Ahmed, J. Albert, M.K. Ketabchi, M.C. Shan, Object Identification in Multidatabase Systems, Proc. 5th IFIP Working Conference on Semantics of Interoperable Database Systems, Lorne, Australia, November 1992.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....its value and location without losing its identity. 6.5. 2 Object Identification Techniques in Multidatabase Systems Object identification techniques within a single database rely on the assumption of a one to one correspondence between real world objects and database objects (proxy objects) [KEN91, KEN93]. This assumption is no longer true in MDB systems, since a real world object may have several corresponding proxy objects with different creation events. Furthermore, these proxy objects may contain conflicting values. Despite this, the end user should be provided with the illusion of a one toone ....

W. Kent et al. Object Identification in Multidatabase Systems. In D. K. Hasio, E. J. Neuhold and R. Sacks-Davis, editors, Interoperable Database Systems (Ds-5) (A-25), pages 313--330, North Holland, 1993. Elsevier Science Publishers B. V.

....object representations (e.g. keys in the relational model or object identifiers in an object oriented model) from two different databases refer to the same object in the world. In this regard, integration mediators build on previous systems that support full [WHW89, WHW90] or partial [ADD 91, KAAK93] materialization for supporting integrated views. In particular, integration mediators can accommodate a variety of complex criteria for matching objects, including look up tables , user defined functions, boolean conditions, historical conditions, and intricate heuristics. The current Squirrel ....

....in different databases correspond to the same object in the world. Most systems that support integration using the virtual approach, including e.g. Multibase and SIMS, assume that a universal key (possibly involving derived attributes) is available for performing object matching. References [KAAK93, WHW90] are among the earliest that consider the problem of object matching in contexts where universal keys are not available. In addition to universal keys, WorldBase [WHW89, WHW90] incorporates look up tables that hold matching information, and introduces negative keys which, intuitively, ....

[Article contains additional citation context not shown here]

W. Kent, R. Ahmed, J. Albert, and M. Ketabchi. Object identification in multidatabase systems. In D. Hsiao, E. Neuhold, and R. Sacks-Davis, editors, Interoperable Database Systems (DS-5) (A-25). Elsevier Science Publishers B. V. (North-Holland), 1993.

.... been added by different systems in order to make the ODMG model feasible as a common data model for heterogeneous DBMS; proposals have been made for adding view mechanisms [7, 1, 17] for object identifiers that do not guarantee immutability during an object s lifetime (so called weak identifiers) [7, 14, 19], and for weak references which are based on weak identifiers [7] Also, a variety of new issues in the area of heterogeneous DBMS have been addressed such as object oriented transaction management [11, 30] and the integration of non traditional DBMS through operational mapping [3] However, a ....

W. Kent, R. Ahmed, J. Albert, M. Ketachi, and S. M-C. Object identification in multidatabase systems. In Interoperable Database Systems (DS-5), (A-25). IFIP, 1993.

....from heterogeneous and geographically distributed information sources, one or more of which may be legacy systems. Among a wide range of techniques addressing this problem, data warehousing, i.e. materializing integrated information in a persistent store, is gaining increasing importance [WHW90, KAAK93, IK93, ZGHW95] This paper describes how we are applying research being developed in the H2O project at the University of Colorado, Boulder, to support data integration. We describe here a broad framework for data integration, that supports both data materialization and virtual view capabilities, ....

....(Spectrum 4) was event triggering by the source databases. 3.1 Materialization This spectrum concerns the approach taken by an integration mediator for physically storing the data held in its integrated view. The choices include fully materialized approach, as presented in references [WHW89, KAAK93] which materializes all relevant information in the data warehouse of the mediator; hybrid approach, as illustrated in the Student Employee example of Section 2, that materializes only part of the relevant information; and fully virtual approach, as presented in [DH84, ACHK93, FRV95] that ....

W. Kent, R. Ahmed, J. Albert, and M. Ketabchi. Object identification in multidatabase systems. In D. Hsiao, E. Neuhold, and R. Sacks-Davis, editors, Interoperable Database Systems (DS-5) (A-25). Elsevier Science Publishers B. V. (North-Holland), 1993.

....and integration of heterogeneous databases. Another problem that must be addressed when integrating schemas is that of inter database object correspondences. Work in this area has focused on interdatabase object reference by description [46] and object identification via language constructs [27]. Based on these earlier results, there are now several efforts focusing on database schema integration, with varied approaches to handling the problem of semantic diversity. One common approach to schema integration is to reason about the meaning and resemblance of heterogeneous objects in terms ....

W. Kent, R. Ahmed, J. Albert, M. Ketabchi, and M. Shan. Object Identification in Multidatabase Systems. Technical report, Hewlett-Packard Laboratories, 1992.

....determining when two object representations (e.g. keys in the relational model or object identifiers in an object oriented model) from two different databases refer to the same object in the world. In this regard, Squirrel mediators build on previous systems that support full [26] or partial [2] [21] materialization for supporting integrated views. In particular, Squirrel mediators can accommodate a variety of complex criteria for matching objects, including lookup tables , user defined functions, boolean conditions, historical conditions, and intricate heuristics. The current Squirrel ....

....to match them. WorldBase [26] and SIMS [3] are two examples using this approach. A generalization of this is to permit keys that involve derived attributes. Lookup table based matching uses a lookup table that holds pairs of immutable OIDs or keys of corresponding objects. References [26] and [21] support lookup tables. Comparison based matching provides in addition the possibility of comparing (possibly derived) attributes of two objects, either with arithmetic and logic comparisons or user defined functions that take the attributes as arguments and return a boolean value, such as the ....

W. Kent, R. Ahmed, J. Albert, and M. Ketabchi. Object identification in multidatabase systems. In D. Hsiao, E. Neuhold, and R. Sacks-Davis, editors, Interoperable Database Systems (DS-5) (A-25). Elsevier Science Publishers B. V. (North-Holland), 1993.

....mappings from both relational and nonrelational (e.g. hypertext, hierarchical, network) systems. Once imported, these schemas can be integrated to form a unified Pegasus schema. A detailed description of the Pegasus system can be found in [2,3] Facilities for schema integration are presented in [15]. This paper describes the technique used in Pegasus to import relational schemas. The importance of translating the schemas of imported external databases to a common data model is discussed in [23] which gives a survey of federated database systems. Descriptions of previous work on data model ....

....and the values in the primary key field(s) of the relation(s) We use the term imported function for functions that are defined on producer types and mapped to the attributes of an external database. A detailed description of the semantics of producer types and imported functions can be found in [15]. The idea of a type whose instances are generated algorithmicly first appeared in the notion of virtual classes and imaginary objects discussed in [1] 2.2. The Relational Import Model The relational schemas which are imported in Pegasus contain declarations of relations, attributes, primary ....

W. Kent, R. Ahmed, J. Albert, M.K. Ketabchi, M.C. Shan, Object Identification in Multidatabase Systems, Proc. 5th IFIP Working Conference on Semantics of Interoperable Database Systems, Lorne, Australia, November 1992.

....if the pair of objects satisfies the predicate. A similar method is the key family identification technique used in WorldBase, where two objects are viewed as the same if they agree on any one of (possibly) several candidate keys [43] Another technique is based on a notion of image functions [19]. Virtual OIDs defined for the external systems are mapped by an image function to virtual OIDs of objects in the (virtual) instance of the integrated schema, and two objects are identified as the same if they map to the same integrated object under the image mapping. The image functions may be ....

.... techniques such as supplying a fixed set of aggregation functions to compute resolvant values, as in Multibase [12, 38] or providing a more general mechanism that an integrator can use to define arbitrary resolution mechanisms, such as the reconciler functions implemented in the Pegasus system [19]. The present paper contains several contributions. First, we show that guaranteeing the transitivity of object identity declarations explicitly is difficult when the integration facilities are used with other standard facilities, such as subtyping, in complex schemas. Second, facilities for ....

[Article contains additional citation context not shown here]

W. Kent, R. Ahmed, J. Albert, M. Ketabchi, and M.-C. Shan. Object identification in multidatabase systems. In Proc. of the IFIP TC2/WG2.6 Conf. on Semantics of Interoperable Database Systems, Lorne, Australia, November 1992.

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