Please use this identifier to cite or link to this item:
Extended Rdf-Chord: in Support of Partial Keyword Queries and Improvement of Indices
近年來，由於P2P系統的廣泛使用，越來越多的資源散布在網路上，因此，如何有效率的找到需要的資源也成為了重要的議題。許多關於這類的研究朝著Peer Data Management System(PDMS)的方向進行，為了能準確的找到所需的資源，PDMS利用語意對資源做描述，並藉由資源描述找尋所需的資源，此方式能夠使得搜尋的正確性大幅提升，並且能夠藉由節省許多不必要的查詢封包以及查詢路徑提升查詢速度。PDMS將資源分布至儲存庫，並可由索引產生與否分為非結構式系統和結構式系統，而結構式系統又由儲存庫分配的方式不同而分成集中式系統和分散式系統。由於非結構式系統在資源數量非常龐大的情況下不能夠保證能找到需要的資源，而集中式系統又容易遭遇單點失敗以及負載過重的問題，因此本計劃選擇朝分散式系統的方向發展。分散式系統使用資源描述產生資源索引，並將索引對映至相對應的網路節點(也就是所謂的「儲存庫」)上。藉由索引，系統可以縮小查詢的範圍以增進效率，然而，索引的使用也造成了維護成本高以及複雜查詢的支援度較低等問題，因此，如何結合資源描述的技術和索引分配的方式以達到支援複雜查詢並降低維護成本的目的將是本計劃所要探討的主題。
In the past decade, due to the widely use of Peer-to-Peer systems, more and more resource are distributed on the Internet. Thus, how to find the needed resources precisely and effectively has become an important issue. Many researchers attempted to resolve this kind of problems by using Peer Data Management System (PDMS). PDMSes utilize semantic metadata to describe shared resources. With these semantic metadata, a PDMS can locate resource accurately, and the performance of query is greatly improved because of the reduction of unnecessary query messages and routing path.There are two types of PDMSes: unstructured systems and structured systems; the unstructured systems broadcast messages to find resources, thus may encounter the query traffic problem and cannot provide search guarantee. The structured systems are classified into centralized systems and decentralized systems. The centralized system may encounter the problem of single point of failure and overloading. Therefore, in this project, decentralized PDMS systems are tackled. The decentralized systems generate resource indices by metadata, and map the indices onto nodes (the set of these mapped nodes is called a “repository”) responsible for the indices. By using the indices, the PDMS can reduce the range of nodes it has to look up and, as a result, the query efficiency is improved. In addition, complex queries can be supported. However, the generation of the indices come with a cost; that is, higher maintenance cost of the P2P system. Therefore, in this project, we attempt to design novel distributed indexing schemes such that complex queries are supported, query efficiency is good, and at the same time, the maintenance cost is reasonably low.
|Appears in Collections:||資訊管理學系|
Show full item record
TAIR Related Article
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.