Web server aims to service clients sensitively and clients wish to explore web sites at fast bandwidth. Nevertheless, sometimes users can only get the web content at a slow response due to the slow communication. Using data buffering technique, web cache provides clients an alternative way to acquire the web content from web server at low cost and high bandwidth. Using web cache technique, users can get fast response at low communication cost. When quantities of data are buffered to the cache server, cache policy becomes an important factor which can clearly affect the performance efficiency. Our contribution is that we take full account of the user's visiting action and redesign the cache policy. Based on this, we compare our policy (UVA) with existing method (LFU) through series of experiments. The results show that our method can efficiently improve cache hit rate.