An evaluation system for string extraction in the airline coupon project

In this paper, we introduce an evaluation system to evaluate the performance of segmentation algorithm aimed to the airline coupon project. It is based on two databases which are constructed by a standard observer method and three error measures. Due to the idea of consolidating subjectivity to objectivity, it is proved effective and objective in segmentation evaluation work in the airline coupon project. It is also convenient to analyse the segmentation results and to flesh out the databases endlessly in a net-based integrated working environment.