Sampling is the process of selecting a subgraph from
an original graph to represent the characteristics of it, at a
given point of time. As real-time mobile streaming networks
are temporal, unbounded and huge to fit in memory it is
difficult to analyze them with a commodity machine. We
need real-time evolving samples that can represent the huge
networks, while maintaining a trade-off between accuracy of
results and cost of computation over huge networks. What
if we have a real-time evolving sample of the stream with
the similar properties and topology of the original graph?
There are a number of algorithms proposed for sampling
of streams [1], [2], [3], and [4] etc. However, there are no
solutions to match all the properties of graphs. If the sample
matches few properties, which sample would yield proper
estimates for directed and weighted evolutionary graphs? In
this section, we refer to the work we carried out in [5].
We implemented three sequential algorithms, space saving
[6], reservoir sampling [1] and a biased random sampling
algorithm [5] to generate sample streams in real time. These
algorithms are briefed below.
Sampling is the process of selecting a subgraph froman original graph to represent the characteristics of it, at agiven point of time. As real-time mobile streaming networksare temporal, unbounded and huge to fit in memory it isdifficult to analyze them with a commodity machine. Weneed real-time evolving samples that can represent the hugenetworks, while maintaining a trade-off between accuracy ofresults and cost of computation over huge networks. Whatif we have a real-time evolving sample of the stream withthe similar properties and topology of the original graph?There are a number of algorithms proposed for samplingof streams [1], [2], [3], and [4] etc. However, there are nosolutions to match all the properties of graphs. If the samplematches few properties, which sample would yield properestimates for directed and weighted evolutionary graphs? Inthis section, we refer to the work we carried out in [5].We implemented three sequential algorithms, space saving[6], reservoir sampling [1] and a biased random samplingalgorithm [5] to generate sample streams in real time. Thesealgorithms are briefed below.
การแปล กรุณารอสักครู่..
