Shared State for Distributed Interactive Data Mining Applications
S. Parthasarathy, and S. Dwarkadas. Distributed and Parallel Databases, (2002)Read again (not completed). Good article. They define some design
goals: clients have different needs to be notified about changes
in the data state, client-controlled memory management, and identifying
what part of the data that has been modified. They have an approach
to the design goal of client controlled coherence. It means that
the client can decide which data update scheme it needs. There are
three similar update schemes that ansures that the client gets the
data only once, when it changes, and at certain time intervals.
The system is called Interact. The articture, figure
1, is that the client maps to a virtual dataspace, which then maps
to the data server. They use association and sequence mining as
applicatins of the Interact system. Association mining is
to discover itemsets that frequently occour together. They also
use sequence mining, which aims to discover events that commonly
occour over a period of time. They call this a temporal database.
They actually use summary structures, which makes this a litte more
to the modelling side than my scope which is more on the primary
modelling side. They seem to to have focused partly on the delivery
of the results to the clients. They use 2-dimansional discretation.
I might have to do the same as it is problematic to make calculations
on higherdimensional structures. It also seems to be diffucult to
visualize higher dimensions than that..
Read again (not completed). Good article. They define some design
goals: clients have different needs to be notified about changes
in the data state, client-controlled memory management, and identifying
what part of the data that has been modified. They have an approach
to the design goal of client controlled coherence. It means that
the client can decide which data update scheme it needs. There are
three similar update schemes that ansures that the client gets the
data only once, when it changes, and at certain time intervals.
The system is called Interact. The articture, figure
1, is that the client maps to a virtual dataspace, which then maps
to the data server. They use association and sequence mining as
applicatins of the Interact system. Association mining is
to discover itemsets that frequently occour together. They also
use sequence mining, which aims to discover events that commonly
occour over a period of time. They call this a temporal database.
They actually use summary structures, which makes this a litte more
to the modelling side than my scope which is more on the primary
modelling side. They seem to to have focused partly on the delivery
of the results to the clients. They use 2-dimansional discretation.
I might have to do the same as it is problematic to make calculations
on higherdimensional structures. It also seems to be diffucult to
visualize higher dimensions than that.
%0 Journal Article
%1 parthasarathy02shared
%A Parthasarathy, Srinivasan
%A Dwarkadas, Sandhya
%D 2002
%J Distributed and Parallel Databases
%K imported
%P 129-155
%T Shared State for Distributed Interactive Data Mining Applications
%V 11
@article{parthasarathy02shared,
added-at = {2008-04-30T12:59:47.000+0200},
author = {Parthasarathy, Srinivasan and Dwarkadas, Sandhya},
biburl = {https://www.bibsonomy.org/bibtex/2db25c5a73196c0d3d20c0ffc9a9764f5/kdubiq},
description = {KDubiq Blueprint},
groupsearch = {0},
interhash = {9ca7bbaf6ab44e7321e3b287fa3a2a7b},
intrahash = {db25c5a73196c0d3d20c0ffc9a9764f5},
journal = {Distributed and Parallel Databases},
keywords = {imported},
note = {Read again (not completed). Good article. They define some design
goals: clients have different needs to be notified about changes
in the data state, client-controlled memory management, and identifying
what part of the data that has been modified. They have an approach
to the design goal of client controlled coherence. It means that
the client can decide which data update scheme it needs. There are
three similar update schemes that ansures that the client gets the
data only once, when it changes, and at certain time intervals.
The system is called \emph{Interact}. The articture, \emph{figure
1}, is that the client maps to a virtual dataspace, which then maps
to the data server. They use association and sequence mining as
applicatins of the \emph{Interact} system. Association mining is
to discover itemsets that frequently occour together. They also
use sequence mining, which aims to discover events that commonly
occour over a period of time. They call this a temporal database.
They actually use summary structures, which makes this a litte more
to the modelling side than my scope which is more on the primary
modelling side. They seem to to have focused partly on the delivery
of the results to the clients. They use 2-dimansional discretation.
I might have to do the same as it is problematic to make calculations
on higherdimensional structures. It also seems to be diffucult to
visualize higher dimensions than that.},
pages = {129-155},
timestamp = {2008-04-30T13:00:23.000+0200},
title = {Shared State for Distributed Interactive Data Mining Applications},
volume = 11,
year = 2002
}