[Purpose/　Significance]　Timely　and　accurate　identification　of　transformative　research　is　of　great　significance　for　exploring　the　research　frontiers　in　science,　as　well　as　facing　the　new　pattern　of　high-quality　development　of　the　country.　[Method/　Process]　Based　on　the　characteristics　of　transformative　research　in　the　dimensions　of　breadth,　intensity　and　speed,　and　by　machine　learning　methods,　the　entropy　weight　method,　factor　analysis　method　and　regression　analysis　method,　the　study　constructed　the　transformative　research　identification　model　from　the　perspective　of　citation.　Applied　the　models　above　to　the　fields　of　gene　targets,　magnetoresistive　effects,　fullerenes　to　evaluate　the　early　recognition　effect　of　different　models,　and　applied　the　model　with　the　highest　early　recognition　effect　to　the　fields　of　molecular　biology,　polymer　science　and　astrophysics　in　2017.　[Results/　Conclusions]　Judging　from　the　“ranking”　and　“top%”　indicators,　the　recognition　effect　of　the　early　recognition　model　of　transformative　research　based　on　regression　analysis　is　better　than　that　of　the　other　two　models,　and　the　identified　transformative　research　ranks　higher,　and　the　transformative　research　could　be　identified　in　the　third　year　after　publication.　The　accuracy　of　the　model　in　the　field　of　gene　targets　and　fullerenes　are　higher　than　in　the　field　of　magnetoresistive　effect.　The　accuracy　and　recall　rate　of　the　early　recognition　model　in　the　transformative　study　based　on　regression　analysis　are　80%,　and　the　average　score　of　the　model　is　0.　764.　©　2024　Editorial　Board　of　Journal　of　Modern　Information.　All　rights　reserved.