For the sake of identifying masquerade attackers in a computer, various alignment algorithms has been proposed. The semi-global alignment algorithm (SGA) is the most effective and also efficient technique to detect these type of attacks till now, but it has not reach the level of accuracy and effectiveness required by large scale and multiuser systems. To support these shortcomings of SGA and to increase both the effectiveness and the performances of this algorithm, the Data-Driven Semi-Global Alignment, DDSGA approach has been proposed. DDSGA has much more improvements and increases the scoring of the systems by adopting different alignment parameters for each single client. Moreover, it accepts small behavior changes in user command sequences by admitting a small changes in low-level representation of the commands functionality. It also do adjustments to changes in the user behavior by modifying the user signature according to its current user behavior in the computer. DDSGA decreases alignment overhead and also parallelizes the detection and the update for better optimization of runtime. The experimental outcomes of this DDSGA alignment show that DDSGA accomplishes a high hit ratio of 88.4 percent, with a low false positive rate of 1.7 percent. It also enhances the hit ratio of the enhanced SGA by about 21.9 percent and minimizes Maxion-Townsend cost by 22.5 percent.