In order to compare our approach (ST and LBLM) with a baseline (BL), we conduct another experiment using the hypothesis testing.
As we described above, ST and LBLM is more suitable for searching Web pages based on the language model.
In this section, we discuss about time complexity ties of the cluster based language model (CBLM)  and LBLM. Especially, we focus on process of obtaining neighboring pages.
In contrast, LBLM can get neighboring pages collected by the hyperlinks in each document.