W. Kim, K. C. Kim, and A. Dale. Indexing techniques for object-oriented databases. In W. Kim and F. H. Lochovsky, editors, Object-Oriented Concepts, Databases, and Applications, pages 371--394, Massachusetts, 1989. Addison Wesley.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....of the spatial objects. For example, consider the intersection of two polyhedra. Besides the need to test all points of one polyhedron against the other, the result of the operation is not always a polyhedron but it may sometimes consist of a set of polyhedra. Object oriented databases SC tree [14], CH tree [14] H tree [16] Temporal databases Monotonic B tree [9] Bitemporal B tree [13] Transaction time [19] Valid time [20] String databases Prefix B tree 15] String B tree [12] Spatial databases location key based [1] UB tree [24] Z ordering [22] High dimensional databases ....

....objects. For example, consider the intersection of two polyhedra. Besides the need to test all points of one polyhedron against the other, the result of the operation is not always a polyhedron but it may sometimes consist of a set of polyhedra. Object oriented databases SC tree [14] CH tree [14], H tree [16] Temporal databases Monotonic B tree [9] Bitemporal B tree [13] Transaction time [19] Valid time [20] String databases Prefix B tree 15] String B tree [12] Spatial databases location key based [1] UB tree [24] Z ordering [22] High dimensional databases iMinMax [21] ....

W. Kim, K.C. Kim, and A. Dale. Indexing tech- niques for object-oriented databases. In ObjectOriented Concepts, Databases, and Applications, pages 371-394. Addison-Wesley, 1989.

....that answering 3 sided queries efficiently is key to solving the problem of indexing classes. Indexing classes is the natural generalization of indexing in the context of objectoriented databases and is very important to their good performance (see [KiL, ZdM] for more information on this area) [KKD, LOL] present solutions to the problem of indexing classes. However, their algorithms are based on heuristics and cannot guarantee good worst case performance. Previous attempts to answer 3 sided queries in secondary memory by implementing priority search trees in secondary memory [IKO, KRV] did not ....

W. Kim, K. C. Kim, and A. Dale, "Indexing Techniques for Object-Oriented Databases," in Object-Oriented Concepts, Databases, and Applications, W. Kim and F. H. Lochovsky, eds., Addison-Wesley, 1989, 371--394.

....simple class attribute (such as the string, integer or boolean) A nested predicate is issued against a nested attribute which contains a reference to an object in the domain of another class. The most intensively studied query type was indexing an inheritance hierarchy against a simple attribute [KKD89, LOL92, KiMo94, KaRa94]. The class hierarchy index (CHindex) KKD89] is based on B trees and maintains a single index for all classes in the inheritance hierarchy. The CHindex efficiently performs match point operations, but is not optimal for range queries. Conversely, H index [LOL92] efficiently performs range ....

....predicate is issued against a nested attribute which contains a reference to an object in the domain of another class. The most intensively studied query type was indexing an inheritance hierarchy against a simple attribute [KKD89, LOL92, KiMo94, KaRa94] The class hierarchy index (CHindex) [KKD89] is based on B trees and maintains a single index for all classes in the inheritance hierarchy. The CHindex efficiently performs match point operations, but is not optimal for range queries. Conversely, H index [LOL92] efficiently performs range operations and reads more pages than CH index ....

Kim, W., Kim, K.C., Dale,A., 1989, Indexing Techniques for Object-Oriented Databases. in Object-Oriented Concepts, Databases, and Applications, W.Kim, F.Lochovsky, eds., Addison-Wesley.

....of a given T are better than others with respect to the resulting query performance. Related work on type hierarchy indexing includes two proposals based on standard B tree technology: the straightforward solution of maintaining one B tree per indexed type (called single class index in [KKD89]) and an approach on replication of OIDs (called Class Division in [RK95] Other proposals extend B trees, e.g. the Class Hierarchy Index [KKD89] maintaining a common B tree for all types and an additional leave node organization scheme. Single type B trees nested according to the ....

.... on standard B tree technology: the straightforward solution of maintaining one B tree per indexed type (called single class index in [KKD89] and an approach on replication of OIDs (called Class Division in [RK95] Other proposals extend B trees, e.g. the Class Hierarchy Index [KKD89] maintaining a common B tree for all types and an additional leave node organization scheme. Single type B trees nested according to the inheritance hierarchy are introduced in [LOL92] as H trees. CG trees [KM94] and hcC trees [SS94] extend B trees with multiple lists to organize ....

Won Kim, Kyung-Chang Kim, and Alfred Dale. Indexing techniques for object-oriented databases. In Won Kim and Frederick H. Lochovsky, editors, Object-Oriented Concepts, Databases, and Applications, pages 371--394. Addison-Wesley, Reading, Massachusetts, 1989.

....describe a representative set of these special methods below. 7.1 Inheritance Hierarchy Access Methods Queries on inheritance hierarchies may be against a particular class in the hierarchy or against the hierarchy tree itself. The main index structures proposed for such queries include CH trees [14] and H trees [15] CH trees are known to perform well when the target of a query is the hierarchy tree itself. On the other hand, H trees perform well when the target of the query is a particular class in the hierarchy. A variation of these trees is the hcC tree [22] hierarchy class chain tree) ....

W. Kim, K. Kim and A. Dale, "Indexing Techniques for Object Oriented Databases", In Object-Oriented Concepts, Databases and Applications, Addison-Wesley (ACM Press), 1989.

....is a more flexible index structure that can be tuned to a given query mix containing both, exact match and range queries. These two reasons led us to the development of the CG tree. The focus of the paper is on introducing the CG tree and on a thorough performance analysis of the CH index [7], the H tree [8, 9] and the CG tree. 1 Introduction Index structures facilitate the fast direct access to large sets of records by some key attribute. Fast means that the number of pages to be read to retrieve the qualifying records is small compared to the total number of pages the records ....

....relationship. Queries in object bases may be evaluated on a single class, i.e. the set of the direct members of the class, or on a class including all its subclasses. Moreover, many query languages (e.g. O 2 SQL [3] or GOMql [5] allow queries to be formulated on arbitrary sets. As shown in [7, 8, 9] the access to the members of several classes by a common key attribute may be supported by a multiple set index structure for these classes. In the context of object bases two index structures based on the B tree for supporting multiple set indexing have been proposed: the class hierarchy ....

[Article contains additional citation context not shown here]

W. Kim, K. C. Kim, and A. Dale. Indexing techniques for object-oriented databases. In ObjectOriented Concepts, Databases, and Applications, pages 371--394, 1989. Addison Wesley.

....Object oriented databases (OODB) have received considerable attention in recent years. One problem cited by practitioners in using OODBs in industry, is their performance. Indexing is a common technique to enhance performance. Two types of indexes special to OODB were described by Bertino and Kim [1, 7, 9]: the Class Hierarchy index and the Nested or Path index. The classhierarchy index provides indexing on some attribute value to a class and its sub classes along the is a (sub super type, generalization) hierarchy. The original class hierarchy indexing scheme does not support well multi class ....

....is used to keep track of set membership. The CH tree is a key grouping scheme since it attempts to store all entries with the same key in one leaf page. Range queries then scan pages which may not be relevant to the query, therefore more pages than necessary are scanned for Range queries. See [7, 9] for a detailed structure of this index. y Note that in Figure 1, the m:n relationship Company City is represented as two REF relationships A Uniform Indexing Scheme for Object Oriented Databases 201 ffl H tree the H tree maintains a separate B tree for every set (class) These B ....

[Article contains additional citation context not shown here]

A. Dale, W. Kim and K. C. Kim. Indexing techniques for object-oriented databases. In W. Kim and F. Luchovsky, editors, Object-Oriented Concepts, Databases, and Applications, ACM Press, pp. 371--394 (1989).

....Thus, the domain of A i (e.g. D j ) partially participates in the join with C i . We justify this case more formally as follows. The distribution of the values of the attribute A i across the classes in the class hierarchy rooted at C can be classified into three 258 Wan Sup Cho et al. cases [14]. Here, we assume that the domain of the attribute A i is D . ffl disjoint: each D object is referred to by one class of C . ffl total inclusive: each D object is referred to by any class in C . ffl partial inclusive: the cases neither disjoint nor total inclusive. Except for the total ....

W. Kim, K. C. Kim, and A. Dale. Indexing techniques for object-oriented databases. In W. Kim and F. Lochovsky, editors, Object-oriented Concepts, Applications, and Databases, pp. 371--394. Addision-Wesley (1989).

....design and all of its components is an example. The method, called Enc, uses B trees to achieve dynamic hierarchical clustering and 1 Each object (except the root) in the hierarchy has exactly one ancestor ( parent object ) on each level. This is not a sub class hierarchy as discussed in [21]. 29 CHAPTER 4. DYNAMIC HIERARCHICAL CLUSTERING 30 encoding to optimize the storage usage. To compare the method with other clustering methods, simulations of variants of clustering methods traditionally used by relational and object oriented databases are done. The simulation benchmark is ....

W. Kim, C. Kim, and A. Dale. Indexing techniques for object-oriented database. In Object-Oriented Concepts, Databases, and Applications, pages 371--394. AddisonWesley, 1989.

....as a static forest hierarchy of classes, is also a special case of external dynamic two dimensional range searching. Together with the different problem of indexing nested objects, as in [26] it constitutes the basis for indexing in object oriented databases. Indexing classes has been examined in [22] and more recently in [25] but the solutions offered there are largely heuristic with poor worst case performance. In Section 3.2, we reduce indexing classes to a special case of external dynamic twodimensional range searching called three sided searching. Three sided range queries are a special ....

....methods for indexing classes are: 1) to build a B tree on the individual extent of each class, and (2) to build a single B tree for all objects by indexing the collection of all objects. We refer to the latter as the single index method. The extensive comparison of these two methods in [22] concludes that the technique of indexing the collection of all objects is better than indexing each class s individual extent separately. The main reason for this is that the technique of indexing individual extents has a very high query time, O(c log B n t=B) that is, the query time increases ....

W. Kim, K. C. Kim & A. Dale, "Indexing Techniques for Object-Oriented Databases," in Object-Oriented Concepts, Databases, and Applications, W. Kim & F. H. Lochovsky, eds., Addison-Wesley, 1989, 371--394.

....predicates can be found in [20, 6, 13, 14, 24, 10, 11, 8] b) For inheritance hierarchy: The access scope of a query against a class may include instances of the class and those of its subclasses. An index structure can support both instances in the same search index. Previous work includes [17, 1, 18, 19, 16, 22, 21], 2. Behavioral For OODB, queries may contain method invocation. The behavioral indexing technique is based on pre computation of method results and storing them into an index. Previous work on behavioral indexing includes [15, 12, 13, 14, 2, 7] The effects of two indexes could be entangled, that ....

....but has been shown by experiments to have better performance than the CH tree. We believe that the integration of the triple node hierarchy with one of these other methods can achieve even better performance. 5. 1 Class Hierarchy Tree The class hierarchy tree (CH tree) was proposed by Kim et al. [17]. It has a simple structure based on the B tree. As subclasses inherit attributes from their superclass, it is possible to maintain an index on a common attribute for all classes in an inheritance sub hierarchy. The domain of an attribute may be either a primitive class or a non primitive ....

K. C. Kim, W. Kim, and A. Dale. Indexing techniques for object-oriented database. In W. Kim and F. H. Lockovsky, editors, object-oriented concepts, Databases, and Applications, pages 371--394. Addison-Wesley, 1989.

....Comp, resulting in the path P er:owns:man:name. Evaluating this path in a naive way by taking an object of the class P er, checking which vehicle he owns and in the case he owns a bus checking if this is manufactured by the Fiat company may be very expensive. Therefore, several indexing techniques [2, 5, 6, 10, 11] and caching techniques [9] have been proposed to accelerate the processing of such paths. In this paper, we address the following problem: given a number of operations on a database schema leading to the same path P and some other database characteristics (such as the number of objects in each ....

....the main principles of the indexing techniques which we will consider. These are the simple index, multi index, inherited index, multi inherited index and nested inherited index. Before starting this discussion we recall some preliminary definitions in the next section which are adopted from [1, 10]. 2.1 Preliminary definitions As already noticed processing a database operation gives rise to the processing of a path of the form P = C 1 :A 1 :A 2 : A n . The following gives a formal definition of a path. Definition 2.1 Given an aggregation hierarchy H , a path P = C 1 :A 1 :A 2 : A ....

[Article contains additional citation context not shown here]

Kim, W., Kim K.C., Dale, A., Indexing Techniques for Object-Oriented Databases, in Object-Oriented Concepts, Databases and Applications, Kim, W., Lochovsky, F., (eds), Addison-Wesley, pp. 371-394 (1989).

....along attribute chains in object bases. The index paths in GemStone [7] are restricted to chains that contain only single valued attributes and their representation is limited to binary partitions of the access path. Similarly, the object oriented access techniques described for the Orion model [6, 1] are extended in several dimensions in our framework. Our technique differs in three major aspects from the two aforementioned approaches: ffl access support relations allow collection valued attributes within the attribute chain ffl access support relations may be maintained in four different ....

W. Kim, K. C. Kim, and A. Dale. Indexing techniques for object-oriented databases. In W. Kim and F. H. Lochovsky, editors, Object-Oriented Concepts, Databases, and Applications, pages 371--394, Addison Wesley, Reading, MA, 1989.

....constitutes the basis for indexing in object oriented databases. Indexing classes has been examined Diagonal corner query Diagonal corner query 3 sided query 2 sided query general 2 dimensional query Figure 1: Diagonal corner queries, 2 sided, 3 sided and general 2 dimensional range queries. in [19], and more recently in [22] but the solutions offered there are largely heuristic with poor worst case performance. In Section 2.2, we reduce indexing classes to a special case of external dynamic 2 dimensional range searching called 3 sided searching. 3 sided range queries are a special case of ....

W. Kim, K. C. Kim, and A. Dale, "Indexing Techniques for Object-Oriented Databases," in Object-Oriented Concepts, Databases, and Applications , W. Kim and F. H. Lochovsky, eds., Addison-Wesley, 1989, 371--394.

.... algorithms to support spatial joins have been developed [3, 14, 19, 26, 32] Another special join algorithm has been developed for joining objects on set valued attributes [18] Another important research area is the development of index structures that allow to accelerate the evaluation of joins [16, 22, 23, 31, 39, 40]. However, if there is no selection prior to a join or the selections exhibit a high selectivity value (i.e. many output tuples are produced) the performance gain of these algorithms is limited. This is also true for bitmap join indices [31] that were developed especially for Data Warehouse ....

W. Kim, K. C. Kim, and A. Dale. Indexing techniques for object-oriented databases. In W. Kim and F. H. Lochovsky, editors, Object-Oriented Concepts, Databases, and Applications, pages 371--394, Massachusetts, 1989. Addison Wesley.

....of object queries. Up to now, a lot of research has been done in object query languages [Cat93] GV92] KKS92] which emphasizes the expressive power of path expressions in user queries. There is also research work done on physical access methods to support path expressions as path indexes [KKD89], Ber91] KM90] has proposed an algorithm for efficiently assembling complex objects. SC89, Gra93] have developed a comparison of traditional value based joins and pointer based joins. System designers often think that in object databases, traditional joins are no longer necessary since objects ....

....operators : Sequential selection scans the whole collection and verifies the predicate for all the objects inside the collection ; for large collection, this is not an efficient operator. Index selection benefits from the existence of indexes on attributes or along a path (path index) [Ber91, KKD89] ; it only accesses the objects satisfying the index predicate. As shown above, if a path expression is not executed using the DFF algorithm, support tables needs to be generated during the execution. A support table memorizes the OIDs of linked objects in different collections. The memorized ....

K.C. Kim, W. Kim and A. Dale. Indexing techniques for object-oriented databases. In W. Kim and F. H. Lochovsky, editors, Object-oriented concepts, Databases , and Applications, 371-392. AddisonWesley, 1989.

.... the sort merge and hash joins see [13, 14] A lot of effort has also been spent on parallelizing join algorithms based on sorting [10, 25, 26, 34] and hashing [6, 12, 36] Another important research area is the development of index structures that allow to accelerate the evaluation of joins [16, 22, 21, 29, 40, 42]. All of these algorithms concentrate on simple join predicates based on the comparison of two atomic values. Predominant is the work on equi joins, i.e. where the join predicate is based on the equality of atomic Permission to copy without fee all or part of this material is granted provided ....

W. Kim, K. C. Kim, and A. Dale. Indexing techniques for object-oriented databases. In W. Kim and F. H. Lochovsky, editors, Object-Oriented Concepts, Databases, and Applications, pages 371--394, Massachusetts, 1989. Addison Wesley.

....for access optimization in object bases. The index paths in GemStone [6] are restricted to chains that contain only single valued attributes and their representation is limited to binary partitions of the access path. Similarly, the object oriented access techniques described for the Orion model [5] are contained as a special case in our framework. Our technique differs in three major aspects from the two aforementioned approaches: ffl access support relations allow collection valued attributes within the attribute chain ffl access relations may be maintained in four different ....

W. Kim, K. C. Kim, and A. Dale. Indexing techniques for object-oriented databases. In W. Kim and F. H. Lochovsky, editors, Object-Oriented Concepts, Databases, and Applications, pages 371--394, Reading, MA, 1989. Addison Wesley.

....referred to by C i objects. Thus, the domain of A i (e.g. D j ) partially participates in the join with C i . We justify this case more formally as follows. The distribution of the values of the attribute A i across the classes in the class hierarchy rooted at C can be classified into three cases [14]. Here, we assume that the domain of the attribute A i is D . ffl disjoint: each D object is referred to by one class of C . C i C i A j D 1 D . 1 D C Fig. 2: A schema graph. 38 Wan Sup Cho et al. ffl total inclusive: each D object is referred to by any class in C . ....

W. Kim, K. C. Kim, and A. Dale. Indexing techniques for object-oriented databases. In W. Kim and F. Lochovsky, editors, Object-oriented Concepts, Applications, and Databases, pp. 371--394. Addision-Wesley (1989).

....techniques have been proposed that Object Vehicle Animal car van cat dog Figure 3: A Class Hierarchy. Class Object f char name; void texture; void shape; void color; list ids; fall the image ids which contain the objectg g; Table 2: An Abstract Object. exploit the class hierarchies [4, 5, 6, 7]. The scope of an objectoriented query can be restricted to a particular class (accessing the shallow extent) for example, targeting at cat only, or expanded to include some or all classes in the hierarchy rooted in that class (accessing the deep extent) targeting at Animal and all its ....

....retrieval of the instances from a single class, but also for efficient retrieval of the instances from a class and all its subclasses in the hierarchy. If the type of the indexing attribute (feature) is a one dimensional vector (integer or character) the h structure is similar to the CH tree [4] ( a directory is built in each leaf node of a B like indexing structure) However, if the type of the indexing attribute (feature) is an n dimensional vector, directories need to be integrated into leaf nodes of a R tree [10] like structure. 2.3 The 2D h Scheme The 2D h scheme provides ....

W. Kim, K. C. Kim, A. Dale, Indexing techniques for Object-Oriented Database, In Object-Oriented Concepts, Databases, and Applications, Addison-Wesley, 371-394, 1989.

....aspects of object queries. Up to now, a lot of research has been done in object query languages [Cat93, GV92, KKS92] which emphasizes the expressive power of path expressions in user queries. There is also research work done on physical access methods to support path expressions as path indexes [KKD89, Ber91, KM90]. KGM91] has proposed an algorithm for efficiently assembling complex objects. SC89] have developed a comparison of traditional value based joins and pointer based joins. System designers often think that in object databases, traditional joins are no longer necessary since objects point at each ....

K.C. Kim, W. Kim and A. Dale. Indexing techniques for object-oriented databases. In W. Kim and F. H. Lochovsky, editors, Object -oriented concepts, Databases, and Applications, 371-392. AddisonWesley, 1989.

.... the sort merge and hash joins see [13, 14] A lot of effort has also been spent on parallelizing join algorithms based on sorting [10, 25, 26, 34] and hashing [6, 12, 36] Another important research area is the development of index structures that allow to accelerate the evaluation of joins [16, 21, 20, 29, 40, 42]. All of these algorithms concentrate on simple join predicates based on the comparison of two atomic values. Predominant is the work on equi joins, i.e. where the join predicate is based on the equality of atomic values. Only a few articles deal with special issues like non equi joins [9] ....

W. Kim, K. C. Kim, and A. Dale. Indexing techniques for object-oriented databases. In W. Kim and F. H. Lochovsky, editors, Object-Oriented Concepts, Databases, and Applications, pages 371--394, Massachusetts, 1989. Addison Wesley.

No context found.

W. Kim, K. C. Kim, and A. Dale. Indexing techniques for object-oriented databases. In W. Kim and F. H. Lochovsky, editors, Object-Oriented Concepts, Databases, and Applications, pages 371--394, Massachusetts, 1989. Addison Wesley.

No context found.

W. Kim, K. C. Kim, and A. Dale. Indexing techniques for object-oriented databases. Technical Report DB-86-006, MCC, 3500 West Balcones Center Drive, Austin, TX 78759, 1987.

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