Database Reference
In-Depth Information
Theorem 1 are not satisfied, so there is a chance to discover missing value of year using the full
merge. Indeed, from P 2 we obtain the answer q ( I 2 ):= ( S 3 , {(“ John ”, “ XML ”, )}). The local answer
q ( I 3 ) is empty. But performing the full merge and using F 3 ( x 2 ), we obtain:
q ( merge (( S 3 , {(“ John ”, “ XML ”, )}), ( S 3 , {(“ Ann ”, “ XML ”, “2005”)})) = = ( S 3 , {(“ John ”, “ XML ”, “2005”}).
Thus, the year of John's publication has been discovered and the using of full merge is justified.
The consequences of Theorem 1 impact also the way of query propagation. The P2P propagation (i.e.
to all partners with the P2P propagation mode ) may be rejected because of avoiding cycles. However,
when the analysis of the query qualifier and TP-XFD's shows that there is a chance to discover missing
values, the peer can decide to propagate the query with the local propagation mode (i.e. the peer expects
only the local answer from a partner, without further propagation) instead of rejecting it. Such an action
can take place in peer P 3 in the case (2) discussed above.
QUERY ANSWERING IN 6P2P
Overall Architecture
6P2P is built around a set of peers having a common architecture and communicating with each other
by sending ( propagating ) queries and returning answers. According to the P2P technology, there is not
any central control over peer's behavior and each peer is autonomous in performing operations, such
as accepting queries, query answering and query propagation. The overall architecture of the system is
depicted in Figure 6.
Each peer in 6P2P has its own local database consisting of two parts: data repository of data avail-
able to other peers, and 6P2P repository of data necessary for performing integration processes (e.g.,
Figure 6. Overall architecture of 6P2P system
Search WWH ::




Custom Search