DefinePK

Abstract

Recent years, the volume of data is increasing rapidly. There is a huge of information available that lead to extremely large datasets. Most of data comes in unstructured forms such as Twitter, Face book, Blogs, and others. Formal Concept Analysis (FCA) is a way to organize data. However, large dataset leads to the complex formal lattice and becomes unreadable. Principal Component Analysis (PCA) using Singular Value Decomposition (SVD) are used to reduce the high dimension of data. This method is able to be used with both fuzzy and crisp formal contexts. In order to select principal components, we combine two rules; first rule is we use Cumulative Explained Variance Fraction and second rule is we examine Cattell’s Scree Graph. This method is compared with other methods using Edit Distance measurement that quantify the distance between original lattice and reduced lattices.