H. Meuss. Logical Tree Matching with Complete Answer Aggregates for Retrieving Structured Documents. PhD thesis, Dept. of Computer Science, University of Munich, 2000.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....XML tree. The index has linear size and is constructible in linear time. To compute the tf and idf for all documents of a certain type we need at most 2q 3 path joins where q is the number of query nodes. The computation of all matches can have an exponential time complexity for pathological cases [Meu00]. However, we show that there is a realistic and practically relevant class of queries and XML documents where we reach an overall time complexity of O(q s) for computing the query result (where s stands for selectivity, i.e. the maximal number of document nodes having the same label) 3 8 ....

H. Meuss. Logical Tree Matching with Complete Answer Aggregates for Retrieving Structured Documents. PhD thesis, Dept. of Computer Science, University of Munich, 2000.

....two matching hotel elements, the second city eight. Complete Answer Aggregates can be computed very efficiently due to their compact (polynomial sized) representation of an exponential number of answers (cf. 6, 7] Queries are evaluated pathwise with the help of a dedicated path index structure [5]. Evaluation is based on a dynamic programming algorithm that guarantees that every database node is touched at most a constant number of times. 3 Exploring Complete Answer Aggregates The design of the user interface of X was guided by the observation that the user needs help in localizing ....

H. Meuss. Logical Tree Matching with Complete Answer Aggregates for Retrieving Structured Documents. PhD thesis, Dept. of Computer Science, University of Munich, 2000.

....For the sake of readability several proofs are omitted in the running text and collected in an appendix. The appendix also contains an algebraic reformulation of our approach that underpins the direct relationship to Kilpel ainen s Tree Matching. The paper represents a summary of the PhD thesis [Meu00] of the rst author. The co author was the supervisor and added many ideas. The paper modi es and extends results presented in [Meu98] In [Meu98] a variant of the concept of a complete answer aggregate had been introduced for a fragment of Tree Matching without left to right ordering ....

....number of free variables. This explains why the use of disjunctions is more dicult. Disjunctions are not part of the tree query syntax introduced above. From the fact that complete answer formulae are AND OR trees it can be seen that speci c disjunctions are compatible with our formalism. See [Meu00] for some suggestions. Most one dimensional models o er specialized query constructs like distance operators (expressing, e.g. that two nodes are direct neighbours or have a distance of not more than n words elements) Horizontal 59 distance operators have not been included in the query ....

[Article contains additional citation context not shown here]

Holger Meuss. Logical Tree Matching with Complete Answer Aggregates for Retrieving Structured Documents. PhD thesis, Dept. of Computer Science, University of Munich, 2000.

....approach. Finally, we review related work in Section 9, and give a conclusion in Section 10. 2 Tree Matching Kilpel ainen introduced in [Kil92] several variations of Tree Matching techniques as simple and intuitive means to answer queries over trees. We decided to use a slightly modi ed version [Meu00] of unordered tree inclusion as the most exible and powerful variant 1 . Figure 1 shows an unordered tree inclusion of a query tree in an XML tree. A match is de ned using a mapping from the query nodes to the nodes of the XML tree that preserves label names and the ancestor descendant ....

H. Meuss. Logical Tree Matching with Complete Answer Aggregates for Retrieving Structured Documents. PhD thesis, Dept. of Computer Science, University of Munich, 2000.

....between distinct answer tuples is no more apparent in their enumeration. In this article, it is rst argued that enumerating answer tuples is often not appropriate and that aggregated answers are preferable. Then, a notion of aggregated answers called Complete Answer Aggregate (CAA) generalizing [25, 24] is introduced and algorithms for computing CAAs are given. We only consider CAAs for semistructured data: In this context, CAAs seem particularly attractive since they re ect the graph structure of the database and the query in a very natural way. It is shown that CAAs enjoy nice complexity ....

....be eciently computed. Note that for many novel applications such as e commerce such a cascade style query answering is needed [18] 8 Related Work Tree Databases: For tree queries and tree databases, a simpli ed form of CAA based on the Tree Matching formalism [20] has already been introduced in [25, 24]. The present paper signi cantly extends over this early work. Query formalisms for semi structured data: Several query models for XML and semistructured data have been designed and or implemented and used [23, 3, 33, 15, 30, 31, 4, 10] cf. 4, 10] for surveys. These query models are more ....

[Article contains additional citation context not shown here]

H. Meuss. Logical Tree Matching with Complete Answer Aggregates for Retrieving Structured Documents. PhD thesis, Dept. of Computer Science, University of Munich, 2000.

....apparent in their enumeration. In this article, it is rst argued that, in the context of semistructured data, enumerating answer tuples is often not appropriate and that aggregated answers are preferable. Then, a notion of aggregated answers called Complete Answer Aggregate (CAA) generalizing [33, 31] is introduced and algorithms for computing CAAs are given. Finally, it is shown that CAAs enjoy nice complexity properties: 1) While the number of answer tuples may be exponential in the size of the query, the size of the CAA is at most linear in the size of the query and quadratic in the size ....

....supporting this form of query answering which can be eciently computed. Note that for many novel applications such as e commerce such a cascade style queryanswering is needed [1] 8 Related Work Tree Databases: For tree databases , a simpli ed form of CAA has already been introduced in [33, 31]. The present paper signi cantly extends over this early work. Query formalisms for semi structures data: Several query models for XML and semistructured data have been designed and or implemented and used [44, 28, 10, 40, 22, 37, 38] cf. 11, 19] for surveys. These query models are more ....

[Article contains additional citation context not shown here]

H. Meuss. Logical Tree Matching with Complete Answer Aggregates for Retrieving Structured Documents. PhD thesis, Dept. of Computer Science, University of Munich, http://www.pms.informatik.unimuenchen. de/publikationen/, 2000.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC