Document searching of Final Project in Polytechnic Harapan Bersama today still displays search results ranked by sequential document matches or commonly called document ranking. Thereby this way causes document data discovery is not clustered on each theme of the final project accurately. Clustering algorithms can be used in categorising or grouping of documents. One of clustering algorithm USAge is by applying the method of K means, a simple algorithm developed by Mac Queen in 1967. From the research that has been done, the document final projects' abstract clustering in Indonesian language by applying the K Means algorithm shows generated a good enough clusters, so it can be recommendation that K-Means clustering method is good enough if applied in retrieval application system, with indicators of distance between clusters produced are very close, that is 0.001 when calculated by the method of Davies Bouldin Index.