S. Chawathe, A. Rajaraman, H. Garcia-Molina and J. Widom, Change detection in Hierarchically structured information, Proc. ACM Int. Conf. Management Data (SIGMOD), Montreal, Quebec, Canada, 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... techniques for versioning have also been proposed by database researchers who have focused on problems such as transaction time management of temporal databases [19] support for versions of CAD artifacts in O O databases [12] and, more recently, change management for semistructured information [7]. In the past, the approaches to versioning taken by database systems and document management systems have often been different, because of the different requirements facing the two application areas. In fact: Database systems are designed to support complex queries, while document management ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina and J. Widom, "Change Detection in Hierarchically Structured Information", Proc. of SIGMOD, 1996.

....of data to materi alize. There is also the problem of knowing when data in a source has changed. For certain sources we may know exactly at what time and with what frequency the source is updated. For other sources we propose to use techniques for change detection such as those developed in [6]. In this paper our focus is on the first factor i.e. the distribution of user queries to decide what classes of data to materialize. In the following section we present an algorithm that determines the frequently accessed classes of data by extracting patterns in user queries. In the first ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom. Change detection in hierarchically structured information. In Proceedings of the ACM International Conference on Management of Data (SIGMOD), Montreal, Quebec, Canada, 1996.

....include information retrieval in linguistic, taxonomic, and neuroanatomical databases, among others. Many algorithms have been developed for tree searching and matching [1, 12, 18, 24] Most of these algorithms focus on comparing two trees based on various distance metrics. Chawathe et al. [7, 8, 9] studied the tree matching problem in the context of change detection for structured and semistructured data. There are also e orts spent in the development of query languages [2, 14, 16, 19, 20] and query processing techniques [10] for trees, with applications to XML and object oriented database ....

S. S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom. Change detection in hierarchically structured information. In Proceedings of the ACM SIGMOD International Conference on Management of Data, pages 493-504, 1996.

....Web sites of interest are periodically consulted, and the difference between the latest version and the previous one is computed as edit scripts for storing and querying multiversion documents. Some of the efficient change detection algorithms are briefly discussed next. Change Detection. LaDiff [11] is a change detection algorithm for semistructured information, and approaches the problem by dividing it into (i)the Good Matching problem and the (ii) Minimum Conforming Edit Script problem. A diff algorithm XyDiff for XML documents is proposed in [28, 23] To match the largest identical parts ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina, J. Widom, "Change Detection in Hierarchically Structured Information", in Proc. ACM SIGMOD, 1996.

....Eine Operation OMOBJ xref= op OMOBJ heit kommutativ, falls OMOBJ xref= comm1 fur alle OMOBJ xref= x und OMOBJ xref= y . There is a large body of work on using the XML tree structure to compute di#erences of XML documents modulo XML equality (see e.g. WDC02] The algorithms (see [CRGMW96] for an introduction) compute partial tree matchings and express these as so called edit scripts that add and delete XML elements and attributes in the source tree to arrive at the target tree. The work has been mainly concerned with finding algorithms for optimal (least cost) edit scripts ....

....similarity The action of changing keys in the data, can lead to un intuitive and computationally sub optimal edit scripts, but does not compromise the method per se. the un keyed case, only the notion of structural isomorphism and of ordered and un ordered trees has been considered e.g. in [CRGMW96] Listing 3. An XUpdate edit script (partly) updating Listing 1 to Listing 4 xu:modifications xmlns:xu= http: www.xmldb.org xupdate xu:variable name= c select= definition CMP[0] OMOBJ[ id= comm1 ] xu:remove select= definition CMP[0] OMOBJ[ id= comm1 ] xref xu:append ....

Sudarshan S. Chawathe, Anand Rajaraman, Hector Garcia-Molina, and Jennifer Widom. Change detection in hierarchically structured information. In ACM SIGMOD International Conference on Management of Data (SIGMOD), pages 493--504, 1996.

....application and vice versa. After the file has been edited (e.g. upon file closing) it is re converted into the intermediate XML document and compared to the downloaded version in order to identify the changes. This is done by the client side FCDP which computes the minimum cost edit script [3, 5] that is transmitted to the FCDP server immediately (when in connected or weakly connected mode) or else upon reconnection. Note that re conversion and change detection can also be a user directed event. On the server, FCDP applies the edit script to the XML version of the original document ....

....In addition, diff based utilities do not recognize hierarchically structured data and are unable to discover movement of data from one location to another. Due to these shortcomings, we are adapting existing research on finding minimum cost edit distances of structured data (see, for example, [5, 23]) This line of research transforms the data into a tree structure. An edit script can change tree A into tree B with a sequence of inserts, deletes and moves of A s nodes so that it looks like B in both shape and content. A minimum cost edit script is one that is least expensive with respect to ....

S. S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom, "Change detection in hierarchically structured information," SIGMOD Record (ACM Special Interest Group on Management of Data), vol. 25, pp. 493--504, 1996.

....then used by our Web information system to store, retrieve, and query multiversion documents. Some of the e#cient change detection algorithms proposed in the literature are briefly discussed next. 2. 1 Change Detection For semistructured information, a change detection algorithm was proposed in [11]. The algorithm, called LaDi#, for hierarchical structures approaches the problem by dividing it into (i)the Good Matching problem and the (ii) Minimum Conforming Edit Script problem. The changes are represented as edit scripts. A di# algorithm XyDi# for XML documents is proposed in [4, 16] At ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina, J. Widom, "Change Detection in Hierarchically Structured Information", In Proc. ACM SIGMOD, 1996.

.... version can be transformed into the other; algorithms to support this computation were proposed in [25, 17] Since we are dealing with XML published relational data, the order of the tuples is immaterial and we can also use the change detection algorithm for semistructured information proposed in [9]. All these algorithms represent the deltas between the documents as edit scripts and return minimum deltas that will transform the old version into the new one. As discussed in [6] for elements that are logically identified by keys, it is semantically preferable to detect changes between ....

.... techniques have also been proposed for databases, often in the context of O O systems and CAD applications [24] The emergence of web information systems and many new web based applications has generated a flurry of interest and research activities, at first focusing on semistructured information [9], and now on XML [13, 25, 14, 6] This interest is due to the fact that (i) traditional version management applications are now migrating to a web based environment [3] ii) there is an increasing realization that e permanence must be achieved and the broken link problem must be fixed [23] and ....

Chawathe, S., Rajaraman, A., Garcia-Molina, H., Widom, J.: Change Detection in Hierarchically Structured Information. Proc. ACM SIGMOD (1996)

.... node u i start(v) start(u) end(v) Figure 1 shows a sample XML document with its durable intervals (for simplicity, the tree level numbers are not shown) The root node is assigned interval [1,2100] while its three children nodes have intervals [10,600] 710,1200] and [1400,2000] Ranges [2,9], 601, 709] 1201,1399] and [2001,2099] remain unused to handle future insertions. 1,2100) A (10,600) C (710,1200) A (1400,2000) D (20,100) A (150,250) A (300,400) A (830,860) A (800,900) D (1000,1100) D (160,200) D (210,240) D (1030,1060) C (1500,1600) A ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina and J. Widom, \Change Detection in Hierarchi- cally Structured Information", Proc. of SIGMOD, 1996.

....and restrictions, but is targeted for situations when external memory is needed to calculate the edit distance. There are several other approaches that allow insertion and deletion of single nodes anywhere within a tree [14 17] Expanding upon these more basic operators, Chawathe, et al. [7] define a move operator that can move a subtree as a single edit operation, and in subsequent work [6] copying (and its inverse, gluing) of subtrees is allowed. These two operations bear some resemblance to the insert subtree and delete subtree operations that are used in this paper, but the ....

....a move operator that can move a subtree as a single edit operation, and in subsequent work [6] copying (and its inverse, gluing) of subtrees is allowed. These two operations bear some resemblance to the insert subtree and delete subtree operations that are used in this paper, but the approaches in [6, 7] are heuristic approaches and the algorithm in [6] operates on unordered trees, making it unsuitable for computing distances between XML documents. 3 Tree Edit Distance Two XML documents produced from the same DTD can have very different sizes on account of optional and repeating elements. Any ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom. Change detection in hierarchically structured information. In Proc. of ACM SIGMOD, pages 493--504, 1996.

....small. 41 General survey on data clustering and classification can be found in [AHD96] Here, we focus on works related to distance metrics for non traditional data types such as trees. The most well known distance measure for trees is Tree Edit Distance and has been extensively studied (e.g. [Tai79, ZS89, CRG96, BCD95, WDC01, CAM02, GJK02, CTZ01]) In tree edit distance problem, a distance between two trees is defined as the summation of the costs needed to convert the source tree to target tree using the pre defined set of edit operations such as insert or delete. Since general tree edit distance problem (also known as Tree to Tree ....

....needed to convert the source tree to target tree using the pre defined set of edit operations such as insert or delete. Since general tree edit distance problem (also known as Tree to Tree editing problem or Tree to Tree Correction problem) for unordered trees is known as NP hard, works in [Tai79, ZS89, CRG96, WDC01] instead try to find an e#cient algorithm for limited cases such as ordered binary trees or trees with additional constraints. The best known algorithm for computing tree edit distance between two ordered trees is by Zhang and Shasha [ZS89] and has the time complexity of roughly O(n ) where ....

[Article contains additional citation context not shown here]

S. S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom. "Change Detection in Hierarchically Structured Information". In ACM SIGMOD, Montreal, Quebec, Canada, Jun. 1996.

.... a node u iff start(v) start(u) end(v) Figure i shows a sample XML document with its durable intervals (for simplicity, the tree level numbers are not shown) The root node is assigned interval [1,2100] while its three children nodes have intervals [10,600] 710,1200] and [1400,2000] Ranges [2,9], 601, 709] 1201,1399] and [2001,2099] remain unused to handle future insertions. Figure 1: A sample XML document. The proposed indexed join algorithms work with either numbering scheme. Note that under both (durable and non durable) schemes, for any two distinct nodes u, v, the following ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina and J. Widom, "Change Detection in Hierarchi- cally Structured Information", Proc. of SIGMOD, 1996.

.... [18] VariousteuscF:q9 for veqO#1#c have also be1 prop bydatabase re seabase who have focuse onprobleF such as transaction time manageme t of te: oraldatabase [19] support for veqjjFq of CAD artifacts in O Odatabase [12] and,more ree tly, change managec t forse19q#qc 9OR# information [7]. Inthe past,the approache to veqj#RFc take bydatabase systes and docume t managec tsysteF have ofte be1 di#e91 t, be cause ofthe di#e7q t reRO#c 9q ts facingthe two applicationareli In fact: Database systes are deese to supportcomple queleF while docume t managec tsyste# are not, ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina and J. Widom, "Change Detection in Hierarchically StructuredInfcturedMT3 f SIGMOD, 1996.

....of new stories on a web page and use a classifier to determine the appropriate position of this story in our internal representation. The various threads in this internal representation serves as input to the classifier. For structural differences, we are building on recent work done at Stanford [3] for measuring change detection in hierarchies of objects. The structural properties we employ are based on the grammatical structure of an HTML page as well as other kinds of metadata associated with a document. Although we describe the specific problem of dealing with HTML pages, it should be ....

S. Chawathe, S. Rajaraman, H. Garcia-Molina, and Widom J. Change detection in hierarchically structured information. In Proceedings ACM SIGMOD Symposium. ACM, 1996.

....practices. A search of scienti c data available on the Web (e.g. 3] shows that archiving is a ubiquitous problem. Even databases of physical constants [16] are less constant than one might naively imagine. A popular approach to keep all versions of data is to use di based technique [24, 5, 10, 19, 20, 6]. A sequence of edit scripts is stored so that one can roll back to any version. There are two problems with this approach. First, as a document goes through many versions, it becomes increasingly costly to recover an old version by undoing the sequence of edit scripts. The second issue is ....

....concentrate on comparisons between our archive and incremental di approach. The experimental results with cumulative di s are shown in [2] In addition to the choice of incremental di s or cumulative di s, we also have a choice of tree di s or line di s. Tree di s have been extensively studied [10, 11, 19, 20, 6] and we used XML Di [7] which is implemented for XML and is downloadable from the Web. However, when compared with line di , XML Di in OMIM Data V1 incremental diffs gzip(V1 incremental diffs) # of versions 0.00 20.00 40.00 60.00 80.00 100.00 SwissProt Data V1 successive diffs ....

[Article contains additional citation context not shown here]

S. S. Chawathe and A. Rajaraman and H. Garcia-Molina and J. Widom. Change Detection in Hierarchically Structured Information. In Proc. of ACM SIGMOD Int'l Conf. on Mgmt. of Data, 1996.

.... for versioning have also been proposed by database researchers who have focused on problems such as transaction time management of temporal databases [13] support for versions of CAD artifacts in object oriented databases [15] and, more recently, change management for semistructured information [3]. In the past, the approaches to versioning taken by database systems and document management systems have often been different, because of the different requirements facing the two application areas. In fact: Database systems are designed to support complex queries, while document management ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina, J. Widom, "Change Detection in Hierarchically Structured Information", In Proc. ACM SIGMOD, 1996.

....can be expressed by a containment mapping. That is, one asks whether the query tree appears, or approximately appears, in a data tree. Here, the approximation is measured by the number of paths in the query tree that do not appear in the data tree [84] or by some other distance functions [22, 23, 24, 85, 96, 104, 107]. The query tree may contain don t cares or wildcards [72] There are xed length don t cares (FLDCs) that may match a single node and variable length don t cares (VLDCs) 84] We shall refer to this class of queries as approximate containment (AC) queries. In general, a query ....

S. S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom. Change detection in hierarchically structured information. In Proceedings of the ACM SIGMOD International Conference on Management of Data, pages 493-504, 1996.

....large segments thereof, is here required. Thus the logical order of the document objects must now be preserved, whereas the order of tuples is immaterial in relational databases. Likewise, the version management techniques proposed for object oriented databases [9] and semistructured information [11] assume that the order between objects is not significant, while this is essential for reconstructing an XML document. Many traditional document versioning systems, such as RCS [13] are edit based. They use edit scripts to represent document changes and to reconstruct different versions ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina, J. Widom, "Change Detection in Hierarchically Structured Information", In Proc. ACM SIGMOD, 1996.

....related to recursive query processing [BMSU86, BR86, DR94] It will be important to learn how best to map model management functions on top of recursion, which is now a part of SQL3. Differencing models: In many cases, differencing models is a lot like differencing graph structures. As shown in [CRGMW96, CGM97] its computational complexity is sensitive to assumptions about the kind of structure that the graph can represent and the available mapping operations. This suggests it will be hard to develop generic algorithms for differencing that are parameterized by the kinds of structures of ....

Sudarshan S. Chawathe, Anand Rajaraman, Hector Garcia-Molina, and Jennifer Widom. Change detection in hierarchically structured information. In Proc. of SIGMOD, 1996.

.... for versioning have also been proposed by database researchers who have focused on problems such as transaction time management of temporal databases [13] support for versions of CAD artifacts in object oriented databases [15] and, more recently, change management for semistructured information [3]. In the past, the approaches to versioning taken by database systems and document management systems have often been different, because of the different requirements facing the two application areas. In fact: ffl Database systems are designed to support complex queries, while document ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina, J. Widom, "Change Detection in Hierarchically Structured Information", In Proc. ACM SIGMOD, 1996.

.... techniques for versioning have also been proposed by database researchers who have focused on problems such as transaction time management of temporal databases [19] support for versions of CAD artifacts in O O databases [12] and, more recently, change management for semistructured information [6]. In the past, the approaches to versioning taken by database systems and document management systems have often been di erent, because of the di erent requirements facing the two application areas. In fact: Database systems are designed to support complex queries, while document management ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina and J. Widom, \Change Detection in Hierarchically Structured Information", Proc. of SIGMOD, 1996.

....to (R; b; a; a 1 ) XMLTreeDi# treats this as grafting a new subtree rooted at b. At the time of writing, neither the source code of the tool nor a description of the operation aggregation method is available. CHAPTER 4. WORK ON SYNCHRONIZATION, MERGING AND DIFFERENCING 35 LaDi# LaDi# [CRGW96] is a structural di#erencing tool for L A T E X documents. In addition to handling updates, inserts and deletes, LaDi# also handles moves. The move operation is built into the LaDi# core algorithm, as opposed to the move operation of XMLTreeDi#. In practice this means that LaDi# identifies ....

....The move operation is built into the LaDi# core algorithm, as opposed to the move operation of XMLTreeDi#. In practice this means that LaDi# identifies moves that XMLTreeDi# would treat as a pair of prune and graft operations. The program is more of a demonstration of the algorithms developed in [CRGW96] than a full blown di# tool: only certain L A T E X elements are supported. However, the main interest in this tool from the point of view of this thesis are the underlying algorithms. The di#erencing algorithm of LaDi# is discussed further in section 4.4. 4.3 Merge tools UNIX di#3 The UNIX ....

[Article contains additional citation context not shown here]

Chawathe S. S, Rajaraman A, Garcia-Molina H. and Widom J."Change detection in hierarchically structured information." In Proceedings of the 1996 ACM SIGMOD International Conference on Management of Data (June 1996, Montreal, Quebec, Canada). pp. 493--504

....XML schema as a tree, some labels of the nodes can be names of the XML tags which carry semantic meaning. A relabel from one node to another semantic unrelated node will cause an undesirable result. Thus the assumption is invalid for the XML domain. We overcome this limitation in our work. LaDi [5] adapts a simple cost model in which insert, delete and move are all unit cost operations, i.e. cost is 1. We now re ne the cost model to take XML characteristics into account. LaDi also assumes that each node of the input trees has a special label that describes its semantics (semantic tag) ....

....may have tags paragraph , section , etc. And for each leaf node in the source tree, there is at most one leaf node in the target tree that is close to it (unique close partner) These assumptions facilitate the matching. MH Di [4] allows exible cost models and drops the assumptions in [5] but then takes quadratic time in the size of the input. There are a number of di erences between the tree matching problem studied in [5, 4] and the speci c problem of matching trees that model XML schemas. First, some of their edit operations such as copy and glue in [4] are not meaningful for ....

[Article contains additional citation context not shown here]

S. S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom. Change detection in hierarchically structured information. In SIGMOD, pages 493-504, 1996.

.... techniques for versioning have also been proposed by database researchers who have focused on problems such as transaction time management of temporal databases [12] support for versions of CAD artifacts in O O databases [14] and, more recently, change management for semistructured information [3]. In the past, the approaches to versioning taken by database systems and document management systems have often been different, because of the different requirements facing the two application areas. In fact: Database systems are designed to support complex queries, while document management ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina, J. Widom, "Change Detection in Hierarchically Structured Information", In Proc. ACM SIGMOD, 1996.

....extensions. The reference based representation to be discussed next, solves these logical problems without performance drawbacks 2.1 The Reference Based Version Model Background. In the following discussion, we describe changes between versions by five commonly used tree edit operations [22] [4] INSERT a subtree, DELETE a subtree, UPDATE the content of a node, COPY a subtree, and MOVE a subtree. We also define that an element is changed if ffl its textual content is updated, or ffl any of its sub element is changed. Otherwise, it is unchanged. A maximum unchanged element is one ....

....of this query was presented in the previous section. ffl Document Evolution History queries. These queries focus on the changes between successive versions of the document. They include queries such as, What s new in version 5 of this document The computation of structured diff discussed in [4, 3] is another example of this query kind. ffl Structural Projection. For instance, if the user requests Section 4.1, Subsection 4.2.2, and Chapter 6 from a document, we can represent this request by the following projection list: PL = 4:1; 4:2:2; 6] In a projection list, we assume that the tree ....

[Article contains additional citation context not shown here]

S. Chawathe, A. Rajaraman, H.G. Molina, J. Widom, "Change Detection in Hierarchically Structured Information", Proc. ACM SIGMOD, 1996, pp. 493-504.

....order among siblings is not significant. For database applications of XML we believe that the unordered tree model is more important. Thus, X Diff is designed to handle unordered tree representations of XML documents. This is the major difference between our work and earlier efforts in this area [4, 5]. High Performance. Change detection on unordered trees is substantially harder than that on ordered trees; 21] has shown it to be NP Complete in general case. By exploiting certain features of XML documents, our algorithm achieves polynomial time in complexity. The remainder of the paper is ....

....differences between flat files. The GNU diff utility is probably the most famous one. This algorithm uses the LCS (Longest Common Subsequence) algorithm [13] to compare two plain text files. CVS, another GNU utility, uses diff to detect differences between two versions of programs [1] In [4], Chawathe et al. pointed out that the techniques employed by these two programs cannot be generalized to handle structured data because they do not understand the hierarchical structure information contained in such data sets. Typical, hierarchically structured data, e.g. SGML and XML, place tags ....

[Article contains additional citation context not shown here]

S. Chawathe, A. Rajaraman, H. Garcia-Molina and J. Widom, "Change Detection in Hierarchically Structured Information", Proceedings of the ACM SIGMOD International Conference on Management of Data, Montreal, June 1996.

....rather than extracting on from scratch. This depends on the change detection algorithms. This so called incremental DataGuide maintenance 13 depends on the ability of the source to notify the DataGuide generation algorithm of changes. Hence, change detection is another active research area [35 42]. E. Myers gave the classic text change detecting and editing algorithms based on the Longest Common Subsequence (LCS) 35] The GNU diff program uses this algorithm [36,37] Change detecting for structured data other than text is more difficult. S. S. Chawathe et al. proposed an algorithm for ....

....editing algorithms based on the Longest Common Subsequence (LCS) 35] The GNU diff program uses this algorithm [36,37] Change detecting for structured data other than text is more difficult. S. S. Chawathe et al. proposed an algorithm for change detection in hierarchically structured information [42]. We have proposed an algorithm for incrementally maintaining DTDs, which faces the similar issues to those, addressed by the DataGuide. Our maintenance approach is discussed in Chapter 7. 3.2 DTD Generators The database group in the University of California at San Diego has conducted some studies ....

[Article contains additional citation context not shown here]

S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom, "Change detection in hierarchically structured information", Proceedings of the ACM SIGMOD International Conference on Management of Data, pp. 493-504, Montreal, Canada, June 1996.

....Active XML GL Rule We use edit scripts for representing the di erence among di erent states of the same document, which can be arbitrarily produced by an editing session or by integrating the document editor with electronic mail. Detecting changes on a document is a general problem, considered in [14, 12, 11]. We assume that the problem can be solved by executing an XML di algorithm, which produces an optional (i.e. unde ned on certain nodes) one to one identity relationship between nodes of the old and new version of the document, so that any two related nodes must be considered as two versions of ....

S. S. Chawathe, A. Rajaraman, H. Garcia-Molina, J. Widom. Change Detection in Hierarchically Structured Information. In 1996 SIGMOD Intl. Conference on Management of Data: pp. 493-504, Montreal, June 1996.

....words and markups within these sentences compared to the total length of the two sentences. With this approach, a sentence may need to be compared with all sentences in the other document. Even though changes to data can be detected, changes to the schema cannot be. Alternatively, Chawathe et al. [7] and Zhang et al. 15, 17, 19] view semi structured documents as trees. Thus, the problem of change detection has been transformed to the problem of finding differences between trees. For any pair of leaf nodes, they either match or not, as determined by LCS. Two internal nodes strictly match if ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom. Change Detection in Hierarchically Structured Information. In Proceedings of the ACM SIGMOD Conference, June 1996.

....object could say whether the domain object o was inserted when going from M b to M 1 . In the literature, the result of a Diff is usually a script that transforms the input model to the output model. The process of constructing the script sometimes involves first producing the mapping, as in [CRGMW96, CGM97] Challenge 7 Is it better to make Diff a separate operation or a specialization of OuterMatch Is there a useful way to express a transformation sequence as a model Can script generation be encapsulated into a useful generic operation 2 5.3 Merge Merging is the activity of moving the ....

....related to recursive query processing [BMSU86, BR86, DR94] It will be important to learn how best to map model management functions on top of recursion, which is now a part of SQL3. Differencing models: In many cases, differencing models is a lot like differencing graph structures. As shown in [CRGMW96, CGM97] its computational complexity is sensitive to assumptions about the kind of structure that the graph can represent and the available mapping operations. This suggests it will be hard to develop generic algorithms for differencing that are parameterized by the kinds of structures of ....

Sudarshan S. Chawathe, Anand Rajaraman, Hector Garcia-Molina, and Jennifer Widom. Change detection in hierarchically structured information. In Proc. of ACM SIGMOD Conf. on Management of Data, 1996.

....classes of data to materialize. There is also the problem of knowing when data in a source has changed. For certain sources we may know exactly at what time and with what frequency the source is updated. For other sources we propose to use techniques for change detection such as those developed in [9]. In this paper our focus is on the first factor i.e. the analysis of the distribution of user queries to decide what classes of data to materialize. In the following section we describe how we extract patterns of frequently accessed classes of data from a user query distribution. We present ....

S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom. Change detection in hierarchically structured information. In Proceedings of the ACM International Conference on Management of Data (SIGMOD), Montreal, Quebec, Canada, 1996.

....replaced with pristine copies obtained from other sites. 2.2 Differences with existing approaches As mentioned in the introduction, there has been a lot of work on tracking changes and failure recovery. For example, a data warehouse monitors data sources and copies changes to a centralized store [1, 21, 29]. In our context, the filesystem is the data source, the reliable archive is the warehouse, and the InfoMonitor is the data extractor. However, unlike data warehousing, the InfoMonitor focuses on the ability to restore data to the original source. Thus the InfoMonitor must save information ....

....the feature set of these reliable systems resemble that of a filesystem. The data store integration problem is a common one in data warehousing. Researchers have investigated the problems of designing the architecture of warehouses, detecting changes in source data and maintaining view consistency [21, 22, 1, 29]. We discuss the relation of data warehouses to our work in Section 2.2. Layered filesystems are examined by Khalidi and Nelson [19] although they assume that filesystems contain hooks to facilitate layering. Archiving data, including legacy data, is a significant challenge facing the digital ....

S.S. Chawathe, A. Rajarman, H. Garcia-Molina, and J. Widom. Change detection in hierarchically structured information. In Proceedings of the 1996 ACM SIGMOD International Conference on Management of Data, June 1996.

....we are particularly interested in the dynamic part of the data. For example, we are interested in finding out which restaurants were recently added, which restaurants were seen as improving, degrading, etc. These changes can be captured by a tool that we have implemented, called htmldiff [9]. The htmldiff program takes two versions of a Web page as input, and produces as output a marked up copy of the Web page that highlights the differences between the two versions based on their semistructured contents. Our htmldiff system allows users to browse the markedup Web page to view the ....

....shown in THEORY AND PRACTICE OF OBJECT SYSTEMS, Vol. 24(4) 1 20 1999 CCC1042 98329 94 020253 18 FIG. 1. Example output from htmldiff Figure 1. The icons (which are in color in the actual output) represent different kinds of change operations: insertions, updates, deletions, etc. For details, see [9]. For reasonably small documents, browsing the marked up HTML files produced by htmldiff to view the changes of interest is a feasible option. However, as documents get larger and changes become more prevalent and varied, one soon feels the need to use queries to directly find changes of interest ....

[Article contains additional citation context not shown here]

S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom. Change detection in hierarchically structured information. In Proceedings of the ACM SIGMOD International Conference on Management of Data, pages 493--504, Montr'eal, Qu'ebec, June 1996.

No context found.

S. Chawathe, A. Rajaraman, H. Garcia-Molina and J. Widom, Change detection in Hierarchically structured information, Proc. ACM Int. Conf. Management Data (SIGMOD), Montreal, Quebec, Canada, 1996.

No context found.

Chawathe, S. S., Rajaraman, A., Garcia-Molina, H., and Widom, J. 1996. Change Detection in Hierarchically Structured Information. In Proceedings of the ACM SIGMOD International Conference on Management of Data (SIGMOD). Montreal, Canada, 493--504.

No context found.

S. S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom. Change detection in hierarchically structured information. In Proc. of the ACM SIGMOD Int. Conf. on Management of Data, pages 493--504, Montreal, Canada, June 1996. ACM Press.

No context found.

S. S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom. Change detection in hierarchically structured information. In ACM SIGMOD International Conference on Management of Data (SIGMOD), pages 493--504, 1996.

No context found.

S. S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom. Change detection in hierarchically structured information. In ACM SIGMOD International Conference on Management of Data (SIGMOD), pages 493--504, 1996.

No context found.

S.S. Chawathe, A. Rajaraman, H. Garcia-Molina, J. Widom, Change Detection in Hierarchically Structured Information, SIGMOD Conf., Montreal, 1996.

No context found.

CHAWATHE S., RAJARAMAN A., GARCIA-MOLINA H., WIDOM J., "Change detection in hierarchically structured information", SIGMOD, vol. 25, num. 2, 1996, p. 493504.

No context found.

S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom. Change detection in hierarchically structured information. In SIGMOD, 1996.

No context found.

S. S. Chawathe et al. Change Detection in Hierarchically Structured Information. In Procs of the Int'l Conf. on Management of Data (SIGMOD'96), pages 493--504, 1996.

No context found.

S. S. Chawathe, A. Rajamaran, H. Garcia-Molina, and J. Widom. Change detection in hierarchically structured information. In Proceedings of the ACM SIGMOD International Conference on the management of Data, pages 493--504, Edinburgh, Scotland, U.K., 1996. 25

No context found.

Cha athe, S., Rajaraman, A., Garcia-Molina, H., Widom, J.: Change Detection in Hierarchically Structured Inf2O ation. Proc. ACM SIGMOD (1996).

No context found.

S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom. Change detection in hierarchically structured information. SIGMOD, 25(2):493--504, 1996.

No context found.

S. Chawathe, A. Rajaraman, H. Garcia-Molina, J. Widom, "Change Detection in Hierarchically Structured Information", in Proc. ACM SIGMOD, 1996.

No context found.

S. Chawathe, A. Rajaraman, H. Garcia-Molina and J. Widom, Change detection in Hierarchically structured information, Proc. ACM Int. Conf. Management Data (SIGMOD), Montreal, Quebec, Canada, 1996.

No context found.

S. Chawathe, A. Rajaraman , H. Garcia-Molina , and J. Widom, "Change Detection in Hierarchically Structured Information," Proceedings of The ACM-SIGMOD `96 Conference, pp. 493-504, Montreal, Canada, June 1996.

No context found.

S. Chawathe, A. Rajaraman, H. Garcia-Molina, and J. Widom, Change Detection in Hierarchically Structured Information, SIGMOD, 1996.

No context found.

CRGMW95, Chawathe S., Rajaraman A., Garcia-Molina H. and Widom J., Change detection in hierarchically structured information. Proceedings of the ACM SIGMOD International Conference on Management of Data, 1996.

First 50 documents  Next 50

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