The Estimation of Rule Measure Based on Principle of Information Diffusion

A continuous data mining based on a session model generates a measure sequence of first-order rule. The parameter estimation for the measure sequence obtains basic characteristic of dynamic evolution, used to explain the interestingness and evolutional regularity of the rule. The information diffusion estimation method for the sequence with a small sample is proposed. Being one of higher order mining technique, it attempts to solve the parameter estimation problem of measure sequence composed of incomplete data set, based on the principle of information diffusion. The algorithms are considered from two aspects of descriptive modeling and predictive modeling, and presented for the diffusion estimation in ascend/descend trend, using the measure sequence regarded as incomplete sample. Experiment results show the effectiveness, fine robustness and simplicity