To　build　a　full　picture　of　previous　studies　on　the　origins　of　SARS-CoV-2　(severe　acute　respiratory　syndrome　coronavirus　2),　this　paper　exploits　an　active　learning-based　approach　to　screen　scholarly　articles　about　the　origins　of　SARS-CoV-2　from　many　scientific　publications.　In　more　detail,　six　seed　articles　were　utilized　to　manually　curate　170　relevant　articles　and　300　nonrelevant　articles.　Then,　an　active　learning-based　approach　with　three　query　strategies　and　three　base　classifiers　is　trained　to　screen　the　articles　about　the　origins　of　SARS-CoV-2.　Extensive　experimental　results　show　that　our　active　learning-based　approach　outperforms　traditional　counterparts,　and　the　uncertain　sampling　query　strategy　performs　best　among　the　three　strategies.　By　manually　checking　the　top　1,000　articles　of　each　base　classifier,　we　ultimately　screened　715　unique　scholarly　articles　to　create　a　publicly　available　peer-reviewed　literature　corpus,　COVID-Origin.　This　indicates　that　our　approach　for　screening　articles　about　the　origins　of　SARS-CoV-2　is　feasible.