How to judge a better skyline algorithm for GIS application
JArod
I just have some draft ideas on this. Any further improvement will be
strongly welcomed!
For GIS application of skyline computation, it is important that one
algorithm can be:
1) efficiency on the cost of CPU and I/O;
Considering the algorithm will be run on the mobile device where the
capability of computation is limited and the data need to be
downloaded from server gradually instead of locating on the local
mobile device.
2) capability of handling update efficiently;
Considering a GPS where we need to update the data with the movement
of the car, the algorithm should be able to update the skyline results
for the updated data with a lower cost.
3) skyline results may be given in an approximate way but every time
it should give out the correct skyline objects.
This means that the algorithm may show the results gradually with the
calculation being processed, but the algorithm should make sure that
one a object is showed as a skyline object, it will be surely a
skyline object but not a candidate which may be removed in the
following processing time.
I don't know whether you are clear about these three points. Please
make any further improvement on these standards.
JArod
igeox
I persue in mobile GIS and Location based sevices(LBS). As you
metioned above, the three topics are intensive and important in the
field. But, for a long time, I have been developing some software on
Mobile GIS according to specific application areas such as Urban
Adiministration, Field Gathering etc. In these fields, efficient
queries and real-time data-update are also the keys. if I have more
time, I would like to discuss on these topics with all our members.