Reflection on Spatial Data Mining – A Brief Review (Perumal et al., 2015)

Before reading this paper, I did not know anything about spatial data mining. This article does a good job of introducing GIS and then how spatial data mining relates to it, but was lacking in its explanations of challenges within the field itself.

Concerning how the article was written, I wish they would have expanded on the issue of big data. The issues and challenges section seemed rather short for the scope of challenges which arise with spatial data mining. Additionally, they did not explain their diagram (Fig.4), which was their big solution to big data issues, at all. More examples in general would have also been helpful.

Concerning the topic of spatial data mining itself, a few topics stood out to me. One is ontologies and how there are no clear cut definitions for computers to understand and follow across all spatial mining projects. It reminded me of the lecture we had on ontologies a few weeks back, and about how tricky it was to pinpoint specific definitions to such general topics. This issue doesn’t mean that spatial data mining is inherently flawed, it just recognizes that the human perspective is still important. It shows that there’s a limit to what we can program computers to do for us.

The other topic that stood out to me was just the entire scope of this process. These algorithms must have so much data running through them. There are also so many factors considered in each of the techniques that the computer must understand and consider. This leaves a lot of potential for error and I wonder how spatial data mining researchers minimize this error.

Comments are closed.