Journal of Digital Information Management K. REN, Y.F. REN 13 6 2015 http://dline.info/fpaper/jdim/v13i6/v13i6_2.pdf Word sense disambiguation (WSD) in biomedical texts is important. The majority of existing research primarily focuses on supervised learning methods and knowledge-based approaches. Implementing these methods requires significant human-annotated corpus, which is not easily obtained. In this paper, we developed an unsupervised system for WSD in biomedical texts. First, we predefine the number of senses for an ambiguous word. Kernel fuzzy C-means clustering is used to group the same sense terms into a set. Each set is mapped to a certain sense to achieve disambiguation. Experimental results on all 50 ambiguous terms from NLMWSD corpus demonstrate that our proposed system outperforms other unsupervised methods. Meanwhile, the kernel fuzzy C-means system is 5% more precise than the state-of-art knowledge-based WSD system on the full NLM-WSD dataset. Our system is highly efficient and accurate for word sense disambiguation in biomedical texts and does not require human-annotated corpus.