Since　Budapest　open　access　initiative　was　launched,　a　large　number　of　full-text　articles　in　the　format　of　XML　are　available,　which　further　promotes　the　technology　management　on　the　basis　of　citation　context　analysis,　such　as　emerging　technology　forecasting,　technology　opportunity　detection　and　innovation　measurement.　Inspired　by　the　success　of　kernel　functions　utilized　to　promote　the　performance　of　SVM　(Support　Vector　Machine)　model,　we　explore　the　potential　of　combining　generative　and　discriminative　models　for　the　task　of　citation　function　and　importance　classification.　In　more　details,　generative　features　are　generated　from　a　topic　model,　Citation　Influence　Model　(CIM),　and　then　fed　to　two　state-of-the-art　discriminative　models,　SVM　and　RF　(Random　Forest),　with　other　13　features　derived　from　citation　contexts　directly　to　identify　important　citations　from　a　brand　new　perspective.　Extensive　experimental　results　on　a　dataset　from　the　Association　for　Computational　Linguistics　anthology　indicate　that　our　approach　outperforms　the　counterparts.　©　2020　ACM.