Abstract: Logs are crucial data for maintenance experts to monitor software systems, and automating log pattern mining can significantly aid in understanding system behavior. However, the interleaving of logs from the same or different tasks can interfere with mining log patterns. To address this issue, a time-weighted iterative log pattern mining algorithm, TSP-Miner, is proposed. This algorithm extracted reasonable time intervals between correlated logs as time weights to accurately identify log correlations and mine log patterns, without requiring task-specific information. By employing a cyclic iteration strategy, TSP-Miner continuously replaced subsequences in a log sequence that match existing patterns. This process simplified the structure of the log sequence, thereby effectively mining log patterns even in the presence of interleaving interference. Experimental results on both real and simulated datasets demonstrate that TSP-Miner outperforms existing algorithms in terms of the quality of mined log patterns.