Akademik Veri Yönetim Sistemi
1st International Conference on Engineering and Applied Sciences Optimization, OPT-i 2014, Kos Island, Yunanistan, 4 - 06 Haziran 2014, ss.1206-1216, (Tam Metin Bildiri)
Document clustering plays an important role in information management and data mining systems. Fast document clustering is required to effectively navigate, summarize and organize large amount of information into a small number of meaningful clusters. This increasing importance of document clustering led to be carrying out the expanded range of methods in this area. The classical clustering algorithms like K-means clustering are frequently referred in applications. However, depending on the recent trends in metaheuristic methods, studies focus on applying these meta-heuristic algorithms to solve clustering problems. Requirement of decreased computing time and the ability to search for global optima makes modern heuristics an effective alternative in document clustering. In this study, a novel meta-heuristic namely imperialist competitive algorithm is applied to the document-clustering problem and the performance of the studied algorithm is benchmarked against particle swarm optimization based clustering method and K-Means clustering. Well-known document datasets in literature are used in experiments and the effect of several different similarity measures are analyzed.