Hello,
In regard to your comments on data fusion vs data integration, I believe that your assertion - while probably the case in many situations - is not universally correct.
In the area of geospatial sciences there is often a desire to combine many different types of data (vector, raster) from different sources (for example: bathymetry/topology, meteorological, habitat, etc) into a single data set which includes all of the points of all of the sets with attributes of various data sets assigned or calculated for locations where data might not exist.
As a concrete example, consider the following: A marine researcher attaches tracking devices to whales so that he can view their migration, dive, mating and feeding behavior. After collecting the data, he desires to combine his collected data with bathymetric data (which might have a totally different mesh), habitat information (which again might have a different mesh), and sonar data revealing krill density in the area of divers (again with a different mesh). Since none of these data is likely to have the same sampling rates or sampling locations, the only logical path is to include all of the points from all of the different data sets as part of the result of the fusion operation. Then appropriate interpolation, attribute assignment, etc can take place at each of the defined points in the fused data set.
In this scenario, the fused data set is not a reduced data set at all and it is not a replacement data set either. The new set is a superset of the points in the input data sets with attributes shared among the entire superset of points.
I would like to explain this in the Wiki entry for data fusion if you do not have any objections.
Please let me know think.
Bradjuhasz (talk) 21:52, 17 December 2008 (UTC)Brad