Unrelated　discrete　opinions　can　be　generally　found　in　social　media-based　data　about　a　certain　topic,　which　cannot　be　well　measured　and　analyzed　using　existing　opinion　dynamics　models.　To　fill　this　gap,　this　study　proposes　a　new　discrete　opinion　dynamics　model　based　on　the　sentiment-　opinion　transformation　mechanism.　In　the　model,　a　matrix　formulating　the　transformation　relationship　between　sentiments　and　opinions　is　considered.　In　addition,　the　model　is　initialized　based　on　an　initial　-sentiment　matrix　of　the　information　instead　of　a　certain　mathematical　distribution.　Subsequently,　we　design　the　simulation　experiments　using　different　values　of　parameters　and　different　network　topologies　to　study　four　effects　on　sentiment　and　opinion　dynamics,　namely,　the　effects　of　the　threshold,　the　effects　of　the　initial-sentiment　matrix,　the　effects　of　the　sentiment-opinion　matrix,　and　the　effects　of　the　second　piece　of　information.　The　results　highlight　the　importance　of　transformation　relation　between　sentiment　and　opinion.(c)　2022　Elsevier　B.V.　All　rights　reserved.