Semantic　focused　crawler　is　an　important　part　of　semantic　vertical　search　engine.　It　is　receiving　increasing　attention　as　a　well　founded　alternative　to　search　web　with　the　problem　of　locating　topical　resource　on　entire　web.　In　order　to　retrieval　documents　related　to　a　given　topic,　in　this　paper,　we　propose　QBLP　Algorithm　which　enable　crawler　adaptive　with　the　changing　environment.　This　feature　makes　it　possible　to　change　behavior　of　focused　crawler　according　to　the　particular　environment　and　its　relationships　with　the　given　input　parameters　during　the　search.　QBLP　Exploited　Q　learning　which　features　whole-life　learning　and　repayment　delay　accompany　with　Bayes　classifier.　It　enables　crawler　to　accumulate　experience　during　the　crawling　and　adjust　strategy　to　achieve　goal　of　making　best　decision　under　any　circumstance.　We　make　a　comparison　among　QBLP,　Best　First　and　Breath　First.　According　to　statistics　from　experiments,　We　find　that　QBLP　is　superior　on　precision　than　others　in　long　time　crawling.　©　2010　IEEE.