Online　course　evaluation　is　critical　to　both　course　selection　and　teaching　effectiveness　for　students　and　teachers.　However,　the　current　online　course　evaluation　methods　have　been　criticized　for　neglecting　learners’　needs　and　their　inefficiency.　Therefore,　a　course　evaluation　framework　based　on　sentiment　analysis　using　machine　learning　is　proposed　in　this　study　to　analyze　a　large　number　of　online　course　review　comments　from　learners.　Initially,　massive　open　online　course　review　comments　were　collected　through　web　crawling.　Then,　sentence-　and　aspect-based　sentiment　analyses　were　performed.　Finally,　a　list　of　aspect　terms　that　reflected　the　learners’　requirements　was　compiled　based　on　the　model-generated　outcomes.　The　model　was　utilized　to　evaluate　an　online　intellectual　property　law　online　course.　Results　demonstrate　that　the　training　models　built　in　this　study　achieve　over　90%　accuracy　and　that　90%–95%　of　learners　are　satisfied　with　the　intellectual　property　law　online　course.　The　learners　are　particularly　satisfied　with　the　teacher’s　teaching　style　and　course　schedule.　However,　the　models　also　highlight　the　insufficient　interactivity　in　the　class　and　the　scarcity　of　novel　course　cases.　The　proposed　framework　provides　a　learner-centric　approach　to　evaluating　online　courses,　thereby　enhancing　the　credibility　of　online　course　evaluation.　This　framework　also　serves　as　a　practical　reference　for　online　course　recommendation　and　construction.　©　2023　by　the　authors　of　this　article.　Published　under　CC-BY.