Text clustering is a significant study field in data mining. In text clustering, hidden-information is revealed by analysing text datasets with data mining techniques. Text clustering is used in many current application areas such as document recognition, document organisation, indexing, and visualisation. Since it is used in important applications it is in need of efficient algorithms. In this paper, k-means based algorithms in literature are investigated and a novel k-means based text clustering algorithm is defined. Proposed algorithm is compared with existing k-means algorithm in literature on five real-world datasets. Obtained implications show that recommended algorithm is efficient and useful.