A General Framework on Temporal Data Mining

Mass processing request has made temporal data mining a vital branch of data mining field. A general framework for temporal knowledge discovery is proposed to define primary concepts in first-order linear temporal logic. The sequence is transformed firstly into linear ordered sequence of events consisted of basic strings. The framework represents a rule in quasi-Horn clause, defines the measures of the first-order formula valuating on a linear state structure, generates the estimator sequence of the measures based on a session model, quantifies the novelty of the discovered rules in terms of deviations among the rules using dynamic time warping distance function, and proves the relevant properties of the concepts. A process model of continuous data mining is developed, based on the session model

[1]  Eamonn J. Keogh,et al.  On the Need for Time Series Data Mining Benchmarks: A Survey and Empirical Demonstration , 2002, Data Mining and Knowledge Discovery.

[2]  Fei Wu,et al.  Knowledge discovery in time-series databases , 2001 .

[3]  Kilian Stoffel,et al.  From Temporal Rules to Temporal Meta-rules , 2004, DaWaK.

[4]  Jian Pei,et al.  From sequential pattern mining to structured pattern mining: A pattern-growth approach , 2004, Journal of Computer Science and Technology.

[5]  John F. Roddick,et al.  Higher order mining , 2008, SKDD.

[6]  Jun-Yi Shen,et al.  Ontology service-based architecture for continuous knowledge discovery , 2005, 2005 International Conference on Machine Learning and Cybernetics.

[7]  John F. Roddick,et al.  A Survey of Temporal Knowledge Discovery Paradigms and Methods , 2002, IEEE Trans. Knowl. Data Eng..