Utility Mining Across Multi-Dimensional Sequences

Knowledge extraction from database is the fundamental task in and data mining community, which has been applied to a wide range of real-world applications situations. Different support-based models, utility-oriented framework integrates utility theory provide more informative useful patterns. Time-dependent sequence are commonly seen real life. Sequence have widely utilized many applications, such as analyzing sequential user behavior on Web, influence maximization, route planning, targeted marketing. Unfortunately, all existing algorithms lose sight fact that processed not only contain rich features (e.g., occur quantity, risk, profit), but also may be associated with multi-dimensional auxiliary information, e.g., transaction can purchaser profile information. In this article, we first formulate problem across sequences, propose novel named MDUS extract <underline>M</underline>ulti-<underline>D</underline>imensional <underline>U</underline>tility-oriented <underline>S</underline>equential To best our knowledge, study incorporates time-dependent sequence-order, quantitative factor, dimension. Two respectively EM SD presented address formulated problem. The former algorithm based transformation, later one performs pattern joins searching method identify desired patterns sequences. Extensive experiments carried six real-life datasets synthetic dataset show proposed effectively efficiently discover knowledge databases. Moreover, better insight, it adaptable situations than current models.