One　major　problem　in　information　retrieval　is　term　mismatch　between　queries　and　documents,　this　paper　proposes　query　analysis　and　refinement　algorithm　for　information　retrieval.　It　extracts　related　terms　from　web　based　corpuses,　then　it　finds　additional　query　term　candidates　by　calculating　the　degrees　of　importance　of　relevant　terms,　at　last,　both　the　terms　are　used　to　retrieval.　This　algorithm　can　improve　the　results　of　search.　Experimental　result　shows　that　the　proposed　algorithm　is　effective.