Based on Forward Reference Object Transaction Identification Algorithm on Web Mining

This paper puts forward a new transaction identification algorithm (TIFRO) in web mining, which is based on forward reference object. Firstly, TIFRO uses the maximal forward reference algorithm (MFR) to get the forward reference path, and then join a final object of the forward reference path to transaction set. Secondly, TIFRO starts a new transaction by judging the reference length (RL) in case of RL is longer than the maximal reference length(TMAX). So TIFRO can find the object set but not the path set of MFR. Thirdly, TIFRO uses SPADE algorithm to discovery frequency patterns. Finally, we compare TIFRO with MFR in performance.