Evaluating web access log mining algorithms: a cognitive approach
Young Woon, Wee Keong Ng, Ee‐Peng Lim · 2005
The wide availability of web access logs makes themideal data sources for the mining of web behavior forelectronic commerce competitiveness. Unfortunately, sincesuch logs were originally meant for debugging purposes,they cannot be used directly for mining. Hence, much workhas been done to preprocess the logs into a suitable formand subsequently mine them. However, such existing techniquesmake various assumptions that are valid only forspecific situations. In addition, there is no fair way to comparethem objectively. In this paper, we design a frameworktermed Web Access Log Mining AlGorithm Evaluator(WALMAGE) to represent electronic commerce scenariosand web access log mining algorithms in a way that facilitatesthe choice of the most appropriate algorithm to use ina particular scenario. We propose a cognitive approach tomodel our framework so that a wide range of user behaviorcan be quantified in order to imbue our framework withrealism and practicality.