Social　media　such　as　Facebook　and　Twitter　has　gained　strong　attention　for　sharing　information,　worldwide　connectivity　and　brand　marketing　all　over　the　globe.　The　prevalent　use　of　social　networking　sites　has　produced　exceptional　amounts　of　data.　The　mining　of　these　social　media　applications　has　its　potential　to　excerpt　illegal　activities　which　may　be　helpful　for　individuals,　business　and　customers.　The　mining　of　data　obtained　from　Facebook　and　Twitter,　can　be　used　to　predict　emotions　of　stakeholder　and　its　analysis　can　provide　very　valuable　information　regarding　behavioral　inclinations　of　the　writer.　The　automatic　sentiment　analysis　to　detect　the　emotional　content　in　textual　data　has　been　widely　used　in　many　research　fields.　Most　of　the　existing　sentiment　analysis　techniques　are　tailored　for　English　Language.　This　paper　presents　Urdu　based　antisocial　behavior　detection　(ASB).　We　aim　in　particular　to　establish　antisocial　behavior　detection　method　and　defining　its　emotional　state.　We　are　intended　to　introduce　a　sentiment　analysis　based　behavioral　model　that　describes　emotions　related　to　antisocial　behavior.　In　addition　to　describing　the　negative　emotional　state,　our　model　will　also　use　the　concept　of　behavioral　tendencies　and　evidence　to　predict　the　possible　behavior　of　social　media　activists　based　on　input　text.　We　will　outline　the　design　of　behavior　detection　systems　based　on　social　media　posts.　The　learning　algorithm　learn　about　emotions　from　social　media　posts　in　the　Roman　Urdu　language　to　predict　user＇s　behavior　regarding　any　specific　post.　The　results　of　this　study　has　verified　that　our　method　has　outperformed　state-of-the-art　methods　in　terms　of　accuracy.　A　bilingual　or　multilingual　ASB　approach　can　be　made　in　future.　©　2020　IEEE.