J. McHugh, et al. "Indexing Semi-structured Data", Technical Report, Computer Science Dept., Stanford University, January 1998.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....feature of XGrind is that it retains the structure of the original XML document in the compressed format also. This means that the compressed document can be parsed using exactly the same techniques that are used for parsing the original XML document. A related major benefit is that XML indexes [12] can be created on the compressed document. Further, updates to the XML document can be directly executed on the compressed version. Lastly, a compressed document can be checked for validity against the compressed version of its DTD. We expect that these properties would be of considerable utility ....

....with its tags and element attribute values replaced by their corresponding encodings. The advantage of doing so is that the variety of efficient techniques available for parsing querying XML documents can also be used to process the compressed document. Second, indexes, such as those proposed in [12], can now be built on the compressed document in similar manner to those built on regular XML documents. Third, updates to the XML document can be directly executed on the compressed version. Finally, a compressed document can be checked for validity against the compressed version of its DTD, ....

J. McHugh, et al. "Indexing Semi-structured Data", Technical Report, Computer Science Dept., Stanford University, January 1998.

....indexing method based on the data model; we give some experimental results about the effectiveness of our index method in Section 5, and then conclude the paper in Section 6. 2. Related works There are many works on indexing techniques in the fields of database [1] XML data management [4] 5] [6] [15] information retrieval [2] 3] 8] 10] 13] and multimedia data management [11] 12] Figure 1 shows related indexing techniques for each specific field. LORE [4] 6] is a famous XML database, which proposes to use value indexing, string indexing, and hubindexing technique to access ....

....6. 2. Related works There are many works on indexing techniques in the fields of database [1] XML data management [4] 5] 6] 15] information retrieval [2] 3] 8] 10] 13] and multimedia data management [11] 12] Figure 1 shows related indexing techniques for each specific field. LORE [4] [6] is a famous XML database, which proposes to use value indexing, string indexing, and hubindexing technique to access semi structured data. These indexing techniques have been successful applied to build indices to support the LOREL query language. But most queries for document retrieval are ....

McHugh, J., Widom, J., Abiteboul, S., Luo, O. and Rajaraman, A. Indexing semistructured data. In Stanford Technical Report, January 1998

.... Figure 1: Path and tree pattern queries Earlier works on index structures use a combination of path summaries, joins between lists of nodes, tree traversal, and value indexes ranging over certain node labels to support a wide range of XML queries, including path and tree pattern queries, e.g. [19, 34, 33] for native XML storage systems, and [12, 10, 29] on top of relational database systems. Various special purpose index structures, e.g. 20, 9] support only certain path queries eciently. Recent approaches to path and tree pattern matching, commonly known as structural or containment joins, ....

....regarding size and other properties of PIDs and our indexes in comparison to the earlier node identi cation and indexing schemes for structural joins. We conclude the paper in Section 6. 2. RELATED WORK The Lore systems represents early work on storing and querying semi structured and XML data [19, 18]. It uses a combination of techniques for query processing. In particular, Lore relies on a DataGuide [21, 13] as a structural summary used to discover path and tree patterns. We utilize a DataGuide, but avoid tree traversal since this becomes inecient when executed over secondary memory [27] ....

J. McHugh, J. Widom, S. Abiteboul, Q. Luo, A. Rajaraman. Indexing semistructured data. Technical report, Stanford University, 1998.

....directions for future work. 2. RELATED WORK In this section we briefly review previous approaches to the problem of indexing structured and semistructured data. Many approaches, which have been mainly proposed by the database community, are close to the research done on semistructured databases [8, 13, 14, 4]. In this case, the main 663 objective is to build semistructured management systems that facilitate query processing. Several query languages have been proposed for this purpose. For an overview, the reader may refer to [6, 1] Other approaches are stemmed by the work done on structured ....

.... of information retrieval [12, 19, 11, 16, 17] Among the database approaches is the Lore system [13] Lore accomplishes the uploading of new documents by adding the elements of the documents in a tree like structure and updating several indexes (value index, text index, link index and path index [14]) From that point on, any data access is performed by considering the whole database as a huge tree that contains XML elements. Lore approach seems to view an XML document as a database and a set of documents as a single large database where all documents are mixed together into a tree like ....

J. McHugh, J. Widom, S. Abiteboul, Q. Luo, and A. Rajaraman. Indexing Semistructured Data. Technical Report, Computer Science Dept., Stanford University (1998)

....3.2 Index storage There is a rich history of work on index structures suited to specific purposes. In particular, we draw inspiration from the work done in the context of object oriented systems, such as [33] More recently, novel path indices have been proposed for XML and semi structured data [41,32,18]. Schema summarization structures have also been proposed [27,28] We are intensively studying this problem, but at the current time have only single node indices implemented in Timber. We construct value indices on attribute values, whether these are numeric or character string. We also ....

J. McHugh, J. Widom, S. Abiteboul, Q. Luo, A. Rajaraman (1988) Indexing semistructured data. Technical Report, January. Available at: http://www-db.stanford.edu/lore/pubs

....databases [4, 5, 13] These approaches do not require consideration of any special index structures, because they rely on the power of relational query processing engine. Actually, variations of this approach are also adopted in commercial DBMSs. Several index structures were defined in Lore [6, 8]. They index values of atomic objects supporting type coercion (Vindex) and keyword search (Tindex) and labeled paths (Pindex) An e#cient index structure Index Fabric was proposed in [3] This structure is well tuned for path selection and is capable of indexing huge amounts of data. However, ....

Jason McHugh, Jennifer Widom, Serge Abiteboul, Quingshan Luo, and Anand Rajaraman. Indexing semistructured data. Technical report, Stanford University, 1998.

....query languages. The first is to build performance enhancing data structures, e.g. indexes, and generate a query evaluation plan utilizing the structures. Lore has several indexes, such as value and path indexes [8] Lorel queries can be compiled into plans that make efficient use of the indexes [9]. Other path indexes include the t index [10] and the Index Fabric [11] For XPath, the Dynamic XML Engine (DXE) takes advantage of available indexes to accelerate queries [12] Warpedge optimization is similar because it builds a path index consisting of the warp Proceedings of EWIS 2002, ....

J. McHugh, J. Widom, S. Abiteboul, Q. Luo, and A. Rajaraman. Indexing Semistructured Data. Technical Report, Stanford University, Database Group, January 1998.

....databases [4, 5, 13] These approaches do not require consideration of any special index structures, because they rely on the power of relational query processing engine. Actually, variations of this approach are also adopted in commercial DBMSs. Several index structures were de ned in Lore [6, 8]. They index values of atomic objects supporting type coercion (Vindex) and keyword search (Tindex) and labeled paths (Pindex) An ecient index structure Index Fabric was proposed in [3] This structure is well tuned for path selection and is capable of indexing huge amounts of data. However, ....

Jason McHugh, Jennifer Widom, Serge Abiteboul, Quingshan Luo, and Anand Rajaraman. Indexing semistructured data. Technical report, Stanford University, 1998.

....3.2 Index Storage There is a rich history of work on index structures suited to specific purposes. In particular, we draw inspiration from the work done in the context of object oriented systems, such as [7,29] More recently, novel path indices have been proposed for XML and semi structured data [15, 28, 37]. Schema summarization structures have also been proposed [23, 24] We are intensively studying this problem, but at the current time have only single node indices implemented in Timber. We construct value indices on attribute values, whether these are numeric or character string. We also ....

J. McHugh, J. Widom, S. Abiteboul, Q. Luo, and A. Rajaraman. Indexing Semistructured Data. Technical Report, January 1998. Available at http://wwwdb. stanford.edu/lore/pubs.

....are local, i.e. DF speci c (tf, df) The other arguments are global, i.e. non DF speci c (dlen, slen) Local and global arguments have to be stored di erently. Global parameters are independent of query terms and can thus be managed within existing structural indexes such as the DataGuide [13]. In most systems, such indexes have to be traversed anyway in order to determine answers for p queries. Thus, the global statistics can be obtained on the y at no additional cost. In the following, we focus on the ecient provision of local weighting algorithm parameters. 3.1 Index Requirements ....

....database system like Natix [10] might serve as a suitable basis for our approach. There is only limited work yet concerned with index structures for and principles of an integrated information retrieval data retrieval index and the kind of supported queries. The Stanford Lore system s Tindexes [13] are text index structures for exact keyword matches used to improve data retrieval. However, these text indexes range only over certain administrator selected elements, not allowing to determine all occurrences of a term. Furthermore, no relevance based queries are supported. In [14, 17] an IR ....

J. McHugh, J. Widom, S. Abiteboul, Q. Luo, A. Rajaraman. Indexing semistructured data. Technical report, Stanford University, Feb. 1998.

....query. Passage retrieval approaches [9] have used more negrained document components to improve the ranking but without supporting IR on document parts as a local context. The feasibility and utility of IR index structures for semistructured data as required by our approach have been shown in [10]. Overview. In Section 2 we establish the background for this paper. In Section 3, we introduce the new operator and demonstrate its application. Section 4 discusses implementation issues of the operator including dynamic ranking, index structures and query optimization. In Section 5, we conclude ....

....associated with a full text index on top of XML data can be compensated by an IR index aware query engine, an advantage that is ignored when deploying IR outside of a database. For semistructured data without a lot of schema information, optimization via the IR index is especially promising [10]. Not only point term indices directly to the data, but detailed data statistics, e.g. about the variance of certain element values are easily integrated into an IR index. 5 Conclusions and Future Work We have introduced a new operator into XQuery that naturally extends XML queries by ....

J. McHugh, J. Widom, S. Abiteboul, Q. Luo, and A. Rajaraman. Indexing Semistructured Data. Technical report, Stanford University, February 1998.

....feature of XGrind is that it retains the structure of the original XML document in the compressed document also. This means that the compressed document can be parsed using exactly the same techniques that are used for parsing the original XML document. A related major benefit is that XML indexes [37] can be created on the compressed document. Further, updates to the XML document can be directly executed on the compressed version. Lastly, a compressed document can be checked for validity against the compressed version of its DTD. We expect that these properties would be of considerable utility ....

....with its tags and element attribute values replaced by their corresponding encodings. The advantage of doing so is that the variety of efficient techniques available for parsing querying XML documents can also be used to process the compressed document. Second, indexes, such as those proposed in [37], can now be built on the compressed document in similar manner to those built on regular XML documents. Third, updates to the XML document can be directly executed on the compressed version. Finally, a compressed document can be checked for validity against the compressed version of its DTD, ....

J. McHugh, J. Widom, S. Abiteboul, Q. Luo and A. Rajaraman, "Indexing Semistructured Data", Technical Report, Stanford University, January 1998.

....to a large index design space and corresponding path evaluation search space, both of which are beyond the scope of this paper. For this work, we consider only one composition: the addition of a label map. The label map is simply a partition of data nodes by label, analogous to the edge index of [19], and allows access to nodes with a certain label. It is implemented as a hash table (We consider it unlikely that any large data graph would be stored without such an mapping, and data nodes may even be stored clustered by label) 4. Path Evaluation with Approximate Index Graphs In this section ....

J. McHugh, J. Widom, S. Abiteboul, Q. Luo, and A. Rajamaran. Indexing semistructured data. Technical report, Stanford Univ., Jan. 1998.

....for Lore s cost based query optimizer [20] They are a top down strategy for exploiting the path expression, a bottom up strategy for exploiting value predicates, and a hybrid strategy. To speed up query processing in a Lore database, four different types of index structures have been proposed [16, 21]. Value index and text index are used to search objects that have specific values; link index and path index provide fast access to parents of an object and all objects reachable via a given labeled path. Keyword search is also important to query XML data, if the structures of XML data are not ....

Jason McHugh, Jennifer Widom, Serge Abiteboul, Qingshan Luo, and Anand Rajaraman. Indexing semistructured data. Technical report, Stanford University, Stanford CA, February 1998.

....that data. One, build a special purpose database system. Example research prototypes are Rufus [31] Lore [21] and Strudel [13] Lotus Notes is an example commercial product[20] Such a system is particularly tailored to store and retrieve XML data, using specially designed structures and indices[23, 24] and particular query optimization techniques[15, 22] To some extent SGML databases [36, 30] or systems like Gras [19] which are designed to store graphs for software engineering environments fall into this category as well. Two, use an object oriented database system. In this approach, the rich ....

J. McHugh, J. Widom, S. Abiteboul, Q. Luo, and A. Rajaraman. Indexing semistructured data. Technical Report, Stanford University, January 1998.

.... examples include the DB2 Text Extender [18] and Oracle InterMedia Text [27] An example of integrating text search with semi structured databases is Lore [23] in which a simplified version of an IR style text index is used to locate strings containing specific text words or groups of text words [22]. None of this previous work explores the performance implications of a special purpose vs. native implementation of this functionality in an RDBMS. The advent of SGML [15] has triggered much research on integrating content and structure in text retrieval, in cluding [3, 38, 4, 30] Work on ....

J.McHugh, J.Widom, S.Abiteboul, Q.Luo, and A.Rajaraman. Indexing semistructured data. In Stanford Technical Report, January 1998.

No context found.

J. McHugh, J. Widom, S. Abiteboul, Q. Luo, and A. Rajaraman. Indexing semistructured data. Technical report, Stanford University Database Group, 1998. Available at ftp://db.stanford.edu/pub/papers/semiindexing98.ps.

No context found.

J. McHugh, et al. "Indexing Semi-structured Data", Technical Report, Computer Science Dept., Stanford University, January 1998.

No context found.

Jason McHugh, Jennifer Widom, Serge Abiteboul, Qingshan Luo, and Anand Rajaraman. Indexing Semistructured Data. Technical report, Stanford University, Stanford, California, 1998.

No context found.

Jason McHugh, Jennifer Widom, Serge Abiteboul, Qingshan Luo, and Anand Rajaraman. Indexing Semistructured Data. Technical report, February 1998.

No context found.

J. McHugh et al. Indexing semistructured data. Technical report, Stanford University, Computer Science Department, 1998.

No context found.

J. McHugh, , J. Widom, S. Abiteboul, Q. Luo, and A. Rajaraman. Indexing semistructured data. Technical report, Stanford University, Stanford, CA, USA, 1998. 156

No context found.

J. McHugh, J. Widom, S. Abiteboul, Q. Luo, and A. Rajamaran. Indexing semistructured data. Technical report, Stanford University, Computer Science Department, 1998.

No context found.

J. McHugh, J. Widom, S. Abiteboul, Q. Luo, and A. Rajamaran. Indexing semistructured data. Technical report, Stanford University, Computer Science Department, 1998.

No context found.

J. McHugh, J. Widom, S. Abiteboul, Q. Luo, and A. Rajaraman. Indexing semistructured data. Technical report, Stanford University, 1998.

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