Document Clustering using Self-Organizing Maps

• Main Authors: Muhammad Rafi, Muhammad Waqar, Hareem Ajaz, Umar Ayub, Muhammad Danish
• Format: Article
• Language: English
• Published: Brno University of Technology 2017-06-01
• Series: Mendel
• Subjects: Document Clustering; Text Mining; Neural Network; Unsupervised Learning; Self-Organizing Maps; Layered Approach
• Online Access: https://mendel-journal.org/index.php/mendel/article/view/61

Cluster analysis of textual documents is a common technique for better ltering, navigation, under- standing and comprehension of the large document collection. Document clustering is an autonomous method that separate out large heterogeneous document collection into smaller more homogeneous sub-collections called clusters. Self-organizing maps (SOM) is a type of arti cial neural network (ANN) that can be used to perform autonomous self-organization of high dimension feature space into low-dimensional projections called maps. It is considered a good method to perform clustering as both requires unsupervised processing. In this paper, we proposed a SOM using multi-layer, multi-feature to cluster documents. The paper implements a SOM using four layers containing lexical terms, phrases and sequences in bottom layers respectively and combining all at the top layers. The documents are processed to extract these features to feed the SOM. The internal weights and interconnections between these layers features(neurons) automatically settle through iterations with a small learning rate to discover the actual clusters. We have performed extensive set of experiments on standard text mining datasets like: NEWS20, Reuters and WebKB with evaluation measures F-Measure and Purity. The evaluation gives encouraging results and outperforms some of the existing approaches. We conclude that SOM with multi-features (lexical terms, phrases and sequences) and multi-layers can be very e ective in producing high quality clusters on large document collections.