Your browser does not support JavaScript!

Home    Δυομότιμο Περιβάλλον Δρομολόγησης και Επεξεργασίας RQL Επερωτήσεων και RVL Όψεων  

Results - Details

Add to Basket
[Add to Basket]
Identifier uch.csd.msc//2005kokkinidis
Title Δυομότιμο Περιβάλλον Δρομολόγησης και Επεξεργασίας RQL Επερωτήσεων και RVL Όψεων
Alternative Title Semantic Query Routing and Processing in P2P Database Systems: The SQPeer Middleware
Creator Kokkinidis, George
Abstract Peer-to-peer (P2P) computing is currently attracting enormous attention. In P2P systems a very large number of autonomous computing nodes (the peers) pool together their resources and rely on each other for data and services. More and more P2P data management systems rely nowadays on intensional (i.e., schema) information for integrating and querying peer bases. Such information can be easily captured by emerging Semantic Web languages such as RDF/S. However, a fully-fledged framework for evaluating semantic queries over peer RDF/S bases (materialized or virtual) is missing. In this thesis, we present the SQPeer middleware for processing RQL queries over peers, whose bases are advertised by RVL views. SQPeer utilizes the notion of RDF/S-based Semantic Overlay Networks for organizing the peers into communities, where semantic queries can be efficiently processed and executed. The notion of RDF/S-based SON is examined with respect to three different architectural alternatives, i.e., a hybrid, a structured and an ad-hoc one, with each one posing different challenges. The novelty of SQPeer lies on the use of intensional peer views used for advertising the content of peer bases. A routing phase is responsible for identifying peer advertisements relevant to a specific query based on appropriate query/view subsumption techniques. On the other hand, a query planning phase uses the obtained routing information to construct appropriate distributed query plans considering data distribution in the system for obtaining both complete and correct results. The produced query plan is executed in a fully distributed way by contacting the necessary peers and sending them the appropriate (sub-)queries. Peer communication for exchanging query (sub-)plans and results relies on appropriate communication channels. Compile and run-time optimization strategies are employed in order to create optimal query plans. Heuristics concerning the ordering of joins and unions are used to transform the query plan into a more efficient equivalent one. Both communication and processing cost is used to decide between data, query or hybrid shipping execution policies. Additionally, adaptability in the formulated query plans is possible by monitoring the query execution and altering the whole or part of the query plan at run-time when peer bases become unavailable or system resources are exhausted. Finally, SQPeer's query processing algorithms allow an interleaved execution of the query routing and planning phases. Especially in a structured P2P setting, this interleaved execution permits the creation of multiple query plans in several iteration steps that when combined produce a complete answer. More importantly, intra-site processing is favored with the additional benefit on obtaining more relevant results as soon as possible.
Issue date 2005-07-01
Date available 2005-07-19
Collection   School/Department--School of Sciences and Engineering--Department of Computer Science--Post-graduate theses
  Type of Work--Post-graduate theses
Views 388

Digital Documents
No preview available

Download document
View document
Views : 3