The main purpose of the proposed algorithm is to efficiently mine erasable patterns considering the weight factor over data stream environments based on the sliding window model. As shown in the results of the performance evaluation section, the efficient data structures and mining and pruning techniques proposed in this paper guaranteed outstanding performance. Meanwhile, let us consider more extreme mining environments such as when very large window and pane sizes are assigned, when the size of transaction data is very large, and when mining requests occur very frequently. In such conditions, it may be difficult to provide real-time mining results to users because of computational overheads. However, such problems can be solved through techniques for real-time stream distributed processing. Since the concept of Big Data emerged because of technological advances in the areas of networks, storage, etc., various works for dealing with Big Data have been studied, and real-time stream processing frameworks such as Apache Storm1 have also been developed. Recall that our mining procedure can independently be processed by a divide-and-conquer manner because the pattern expansion works can be performed independently for each item of the header table of the proposed tree structure. Hence, distributed processing techniques like MapReduce can effectively be applied into the implementation of the proposed algorithm. In addition, the proposed architecture and sliding window techniques can also be applied into the stream processing framework because they were designed to be suitable for stream pattern mining. In a future work, we are scheduled to conduct research for applying the techniques and architecture of our algorithm to popular distributed stream processing frameworks.