Some reflections on the clustering algorithm

I can summarize the papers I have been reading lately in three categories:

1. Research around the SOM method (Self-Organising Maps): SOM and WEBSOM are efficient algorithms to find correspondence between textual documents and to map these similarities on a two-dimensional graphical representation.

2. Research around the context automatic recognition: studies of Flanagan, reports and details methodologies for  aggregating sensory data into arrays and matrices for unassisted clustering.

3. Research around spatial data mining: a canadian research group directed by Han has detailed methodologies like CLARANS for clustering elements with no geographical features using their geographical position.

This last group, in fact, results particularly useful for my research. What I need at this point is to put together methodologies for semantic and geographic clustering in some ways. Is there a priority between these two? Can we define a mixed approach?  Are these two techniques independent? How can I merge them?

Tags: , ,

Leave a Reply