Aiming　at　the　personalization　of　recommendation　systems　in　Web　Intelligence,　this　paper　presents　a　new　commodity　recommendation　method　based　on　Bayesian　networks,　which　includes　two　phases:　the　building　of　customer　shopping　model　of　Bayesian　network　and　the　generation　of　recommendation　sets　based　on　probability　inference.　Experimental　results　on　real　world　data　show　that　this　method　is　an　effective　method　that　can　achieve　personalized　recommendation　for　different　customers.