2026-06-25T10:41:29Z https://meral.edu.mm/oai

oai:meral.edu.mm:recid/3116 2021-12-13T05:47:05Z 1582963413512:1596119372420 user-ytu

Modified K-Means for Document Clustering System Tin Thu Zar Win Moe Moe Aye <p>In&nbsp; today&rsquo;s&nbsp; era&nbsp; of&nbsp; World&nbsp; Wide&nbsp; Web,&nbsp; there&nbsp; is&nbsp; a<br> tremendous&nbsp; proliferation&nbsp; in&nbsp; the&nbsp; amount&nbsp; of&nbsp; digitized&nbsp; text<br> documents. As there is huge collection of documents on the web,<br> there&nbsp; is&nbsp; a&nbsp; need&nbsp; of&nbsp; grouping&nbsp; the&nbsp; set&nbsp; of&nbsp; documents&nbsp; into&nbsp; clusters.<br> Document&nbsp; clustering&nbsp; plays&nbsp; an&nbsp; important&nbsp; role&nbsp; in&nbsp; effectively<br> navigating&nbsp; and&nbsp; organizing&nbsp; the&nbsp; documents.&nbsp; K-Means&nbsp; clustering<br> algorithm&nbsp; is&nbsp; the&nbsp; most&nbsp; commonly&nbsp; document&nbsp; clustering&nbsp; algorithm<br> because it can be easily implemented and is the most efficient one<br> in&nbsp; terms&nbsp; of&nbsp; execution&nbsp; times.&nbsp; The&nbsp; major&nbsp; problem&nbsp; with&nbsp; this<br> algorithm is that it is quite sensitive to selection of initial cluster<br> centroids. The algorithm takes the initial cluster center arbitrarily<br> so it does not always promise good clustering results. If the initial<br> centroids&nbsp; are&nbsp; incorrectly&nbsp; determined,&nbsp; the&nbsp; remaining&nbsp; data&nbsp; points<br> with the same similarity scores may fall into the different clusters<br> instead of the same cluster. To overcome this problem,&nbsp;&nbsp; modified<br> K-Means&nbsp; approach&nbsp; is&nbsp; proposed&nbsp; to&nbsp; improve&nbsp; the&nbsp; quality&nbsp; of<br> clustering&nbsp; in&nbsp; this&nbsp; paper.&nbsp;&nbsp;&nbsp; Unlike&nbsp; the&nbsp; traditional&nbsp; K-Means<br> clustering, the proposed K-Means method can generate the most<br> compact and stable clustering results based on maximum distance<br> initial centroids points instead of random initial centroid points.<br> Moreover,&nbsp; the&nbsp; similar&nbsp; data&nbsp; points&nbsp; are&nbsp; clustered&nbsp; based&nbsp; on<br> maximum probability distribution of data points.&nbsp; Therefore, the<br> proposed method is more effective and converges to more accurate<br> clusters than original K-Means clustering method. In this paper,<br> experimental&nbsp; results&nbsp; are&nbsp; presented&nbsp; in&nbsp; F-measure&nbsp; using&nbsp; 20-News<br> Group standard dataset.</p> 2016-10-01 http://hdl.handle.net/20.500.12678/0000003116 https://meral.edu.mm/records/3116