基于MapReduce的可扩展协同聚类算法

协同聚类算法是通过同时对文档和特征进行聚类的一种聚类算法，该算法可以挖掘文档内部特征之间的潜在关系从而达到提高聚类效果的目的。随着大数据时代的到来，算法的并行化显示出它的优越性，为此本文对协同聚类算法进行全面的研究，并扩展它的并行算法，研究基于最小化残差平方和的协同聚类算法，利用MapReduce模式设计与实现协同聚类算法的并行化。实验结果表明，本文提出的并行协同聚类算法能够提高聚类的效率，并具有很好的可扩展性。

A Scalable Co-clustering Algorithm Based on MapReduce

Collaborative clustering algorithm is a kind of clustering algorithm to cluster the documents and the features at the same time, this algorithm can find the potential relationship between internal document features so as to improve the clustering effect. With the arrival of the era of big data, parallel algorithm showed its superiority, this paper carries out a comprehensive research on collaborative clustering algorithm, and extends the parallel algorithm of it. We studied the collaborative clustering algorithm based on minimum sum-squared residue, and then designed and realized the parallel collaborative clustering algorithm with MpReduce model. Experimental results show that the proposed parallel collaborative clustering algorithm can improve the efficiency of clustering, and be of well scalability.