Concept Based Lattice Mining (CBLM) Using Formal Concept Analysis (FCA) for Text Mining

Extracting relevant resources according to a query is imperative due to the factors of time and accuracy. This study proposes a model that enables query matching using output lattices from Formal Concept Analysis (FCA) tool, based on Graph Theory. The deployment of FCA concept lattices ensures that the matching is done based on extracted concepts; not just mere keywords matching hence producing more relevant results. The focus of this study is on the method of Concept Based Lattice Mining (CBLM) where similarities among output lattices will be compared using their normalized adjacency matrices, utilizing a distance measure technique. The corresponding trace values obtained determines the degree of similarities among the lattices. An algorithm for CBLM is proposed and preliminary experimentation demonstrated promising results where lattices that are more similar have smaller trace values while higher trace values indicates greater dissimilarities among the lattices.