With the growing interest in commercial trend analysis, mining association rules in large databases has been an important topic in data mining field. Mining inter-transaction association rules creates new challenges since it involves data items in multiple transactions. In this paper, we study the problems of mining various patterns from large databases and classify them into three types. Moreover, we show the relationships between these problems and illustrate how to transform one into another. After that, we propose a graph-based approach to discover large inter-transaction itemsets. The proposed approach is good at finding the large inter-transaction k-itemsets for a specific value of k and the maximal large inter-transaction itemsets. The experimental results show that our approach performs well in these aspects.