Authors present a new approach to the clustering of descriptive-textual geological data with the aim to identify geological objects. Data used for the research come from the description of boreholes drilled in Lithuania and representing the Baltic Silurian basin. The problems under study include data preparation for analysis, methods of evaluating the number of clusters for Silurian data clustering, attribute elimination, techniques for data clustering and uncertainty elimination. A comprehensive description of clustering results is presented by an expert of geology.
The proposed method of clustering is of general purpose and can be applied to data that come also from any other domain alongside geology. Clustering results and their interpretation presented in the paper can be used for the Lithuanian Silurian overview only, because the identified rock types (clusters) strongly depend on the properties of the primary dataset. Those properties include subjective nature, multiple authorship and the unequal level of detailing.

Issues:

2011 - Vol.53
No. 1, No. 2, No. 3

2010 - Vol.52
No. 1-4

2009 - Vol.51
No. 1-2, No. 3-4

2008 - Vol.50
No. 1, No. 2, No. 3, No. 4, No. Priedas

2007
No. 1, No. 2, No. 3, No. 4

2006
No. 1, No. 2, No. 3, No. 4

2005
No. 1, No. 2, No. 3, No. 4

2004
No. 1, No. 2, No. 3, No. 4

2003
No. 1, No. 2, No. 3, No. 4

2002
No. 1, No. 2, No. 3, No. 4

2001
No. 1, No. 2, No. 3, No. 4