Data mining focuses on discovering interesting, non-trivial and meaningful information from large datasets. Data clustering is one of the unsupervised and descriptive data mining task which group data based on similarity features and physically stored together. As a partitioning clustering method, K-means is widely used due to its simplicity and easiness of implementation. But this method has limitations such as local optimal convergence and initial point sensibility. Due to these impediments, nature inspired Swarm based algorithms such as Artificial Bee Colony Algorithm, Ant Colony Optimization, Firefly Algorithm, Bat Algorithm and etc. are used for data clustering to cope with larger datasets with lack and inconsistency of data. In some cases, those algorithms are used with traditional approaches such as K-means as hybrid approaches to produce better results. This paper reviews the performances of these new approaches and compares which is best for certain problematic situation.

翻译：数据挖掘旨在从大规模数据集中发现有趣、非平凡且有意义的 信息。数据聚类是一种无监督的描述性数据挖掘任务，它基于相似性特征将数据分组并物理存储在一起。作为一种划分聚类方法，K-means因其简单易实现而被广泛使用，但该方法存在局部最优收敛和初始点敏感等局限性。由于这些障碍，受自然启发的群体算法（如人工蜂群算法、蚁群优化算法、萤火虫算法、蝙蝠算法等）被用于数据聚类，以处理更大规模、数据缺失且不一致的数据集。在某些情况下，这些算法与传统方法（如K-means）结合形成混合方法，从而产生更好的结果。本文综述了这些新方法的性能，并比较了它们在特定问题情境下的最佳适用性。