Information Technology Reference
In-Depth Information
actual music piece is required in order to compare
its content with the content of the music pieces
already available in a database.
As with regard to the infrastructure for ex-
changing music data, peer-to-peer networks over
the WWW have gain significant popularity. A
peer-to-peer (P2P) network is a distributed system
in which peers employ resources that are distrib-
uted in the network of peers in order to perform a
function in a decentralized manner. Nodes in P2P
networks normally hold equivalent roles, thus, also
called peers. Within the advantageous qualities
of the P2P networks lie the increased size of the
overall database offered by a P2P network, its fault
tolerance support to peer failure by other peers
and the workload distribution over a network of
available CPUs, since CBMIR is computationally
highly intensive, the absence of the requirement
of special administration or financial arrange-
ments and their self-organzation capability and
adaptability. Additionally, P2P networks offer
the ability to harness the collaborative efforts of
users to provide various semantic tags aiming
at musical content description. Nonetheless, the
very advantages of the P2P network are the same
parameters that make P2P information retrieval
much more complex than in the traditional server-
client model. That is, the lack of a central reposi-
tory for the documents to be retrieved, the large
number of documents available and the dynamic
character of the network, introduce an increased
degree of difficulty in the retrieval process. Ac-
cordingly, as collections become larger, CBMIR
in P2P networks presents new and challenging
requirements the highlights of which are:
Though, despite the previously mentioned
advantages of P2P networks, the trend of musical
data dissemination over P2P networks became
obscure by the illegal exchange of copyrighted
material. One of the key advantages of P2P
networks, as previously discussed, is the lack of
necessity for an administration, which acted as
a loophole. Accordingly, numerous approaches
(Chu, Su, Prabhu, Gadh, Kurup, Sridhar et al.,
2006; Kalker, Epema, Hartel, Lagendijk, & Steen,
2004; Praveen, Sridhar, Sridhar, & Gadh, 2005)
for the protection and reproduction of intellectual
property have been proposed, while the field is
still developing (Dhamija & Wallenberg, 2003;
Dubosson-Torbay, Pigneur, & Usunier, 2004;
Sastry, 2005; Schyndel, 2005), both in terms of
technology as well as ethics. CBMIR applications
in P2P networks can, and must, adopt any such
developments.
The field of music information retrieval has
received increased attention during the last de-
cade. Numerous surveys examine the state of the
art developments in the area (Byrd & Crawford,
2002; Karydis, Nanopoulos, & Manolopoulos,
2006; Orio, 2006; Typke, Wiering, & Veltkamp,
2005) while a litany of works spawns rapidly in
all directions of MIR.
Although outside the scope of this chapter,
work in multimedia (other than music) information
retrieval in P2P networks (Lew, Sebe, Djeraba,
& Jain, 2006) shows a wealth of research issues
that still remain open. Moreover, multimedia IR
in P2P faces similar problems (such as indexing
high-dimensionality data) appearing to MIR in
P2P, though the difference of the nature of the
data, such as interfaces, data representation as well
as data volume issues, requires, in many cases, a
completely differentiated approach. Additionally,
multimedia IR research is naturally complemented
by streaming in P2P networks research as many
of the retrieved multimedia documents are vid-
eos. Therein lie more open research issues (Liu,
Kundur, Merabti, & Yu, 2006).
•
Richer set of search semantics that can sup-
port efficient CBMIR
•
Appropriate P2P models that ensure scal-
ability
•
Distribution of the workload over a network
of available CPUs, as CBMIR is computa-
tionally intensive.
Search WWH ::
Custom Search