Arguably　the　rapid　development　of　Internet　financial　is　one　of　the　most　significant　breakthroughs　in　the　financial　domain.　Automated　financial　statistics　have　gradually　substituted　the　traditional　manual　statistical　methods,　providing　a　reliable　data　basis　for　economic　planning.　Therefore,　the　quality　of　a　business　activity　heavily　relies　on　the　accuracy　analysis　of　user　preferences　and　recommend　rated　products　to　the　users.　Traditional　item-based　collaborative　filtering　method　plays　a　dominant　role　for　analyzing　user　preference　and　recommending　the　items　for　users,　this　method　mainly　utilize　the　fully　rating　data　to　predict　whether　the　user　like　the　target　item.　However,　in　many　cases,　the　available　user　rating　data　is　sparsely,　which　makes　traditional　item-based　collaborative　filtering　method　inefficient　and　inapplicable.　To　address　this　problem,　this　paper　propose　an　ontology-based　user　preference　statistical　model　(ontology-based　UPS),　where　the　concept　and　attribute　features　are　extracted　from　financial　ontology　for　semantic　similarity　computing;　later,　it　is　combined　with　the　calculated　rating　similarities　to　improve　the　accuracy　of　the　similar　item　set　for　the　target　item.　The　research　results　show　that　our　approach　outperformed　traditional　collaborative　filtering　method.　©　2020,　Springer　Nature　Singapore　Pte　Ltd.