|
|
| Identifier |
uch.csd.msc//2006chrysakis |
| Title |
Επεξεργασία κ-κορυφαίων ερωτήσεων σε ομότιμα δίκτυα |
| Alternative Title |
Τop-k Query Processing In schema-based peer-to-peer networks |
| Creator |
Chrysakis, Ioannis C
|
| Abstract |
The idea of Peer-to-peer (P2P) computing offers new opportunities for building highly distributed data systems. As the idea of Semantic Web arisen a new category of peer-to-peer systems called Schema-Based come up. In Schema-Based P2P systems each peer is a whole database management system in itself. Each peer can use its own database schema, manages its own data and by this way keeps sovereignty over itself. Considering a Schema-Based peer-to-peer network our main goal is the easy sharing of knowledge bases which implies efficient exchange of data across the p2p network without consuming enough bandwidth. For this reason at first we suggest a suitable peer-to-peer architecture and a well defined query routing context. Our main contribution is the suggestion of a query routing strategy and a query processing strategy. The proposed query routing strategy directs the query only to a set of relevant peers in such way to avoid network traffic and bandwidth consumption. Our processing technique based on the idea of top-k queries that arisen from the research area of databases. Simply top-k queries return only the k best results according to a given criterion. Recently top-k retrieval algorithms for distributed networks have been presented at some approaches. After presenting these approaches and determining their advantages and drawbacks, we finally conclude the Hybrid Threshold (HT) algorithm could be the best solution for top-k processing in peer-to-peer networks. We extend HT and adapt it under our well defined peer-to-peer environment and in consequence we suggest two improved versions: HT-p2p and HT-p2p+. The first assumes that results are returned by executing an instance of the algorithm to a specified Super-Peer named at this case collector Super-Peer. The last assumes that results come from the combination of all top-k object set that are returned from each running instance of the algorithm to each specified contributor Super-Peer. In addition, since HT-p2p belong to score-based top-k algorithms so we study the problem of scoring objects and suggest accordingly three use cases of the algorithm. For the evaluation of HT-p2p and HT-p2p+ we implement a prototype system built upon JXTA platform. The results of the experiments upon HT-p2p system showed that our suggested algorithm is a scalable, and efficient enough top-k processing algorithm that could be used by any Super-Peer based peer-to-peer network.
|
| Issue date |
2006-04-01 |
| Date available |
2006-07-19 |
| Collection
|
School/Department--School of Sciences and Engineering--Department of Computer Science--Post-graduate theses
|
|
|
Type of Work--Post-graduate theses
|
| Views |
402 |
Επεξεργασία κ-κορυφαίων ερωτήσεων σε ομότιμα δίκτυα
