Abstract: Traditional algorithms for frequent subgraphs mining have limits when dealing with biological datasets.Biological network has its own characters.Based on these characters，authors propose a new algorithm considering imbalance across datasets，called IFS（Iterated Function System），for mining frequent subgaraphs by relative support.Through dealing with the real protein interaction networks，it is proved that the algorithm is feasible.

Key words: imbalance, vertex support, relative support

摘要： 传统的图挖掘算法应用到生物数据上有其局限性。根据生物网络的特性，通过引入相对支持度的概念，提出了一种解决数据样本间不平衡性的频繁子图挖掘算法——IFS算法。通过对真实的蛋白质互作网络进行处理，证明该算法是可行的。

关键词: 不平衡性, 节点支持度, 相对支持度