Given　the　urgent,　dynamic,　and　uncertain　nature　of　emergencies,　the　rapid　response　to　emergencies　and　their　control　is　challenging,　with　the　intelligent,　real-time　acquisition　of　decision　data　being　a　key　issue.　However,　most　previous　studies　relied　on　simulations,　experts,　assumptions,　questionnaires,　and　borrowed　literature,　which　are　impractical.　Therefore,　this　study　developed　two　methods　for　the　intelligent　and　objective　acquisition　of　quantitative　decision　data　based　on　machine　learning　and　online　review　data　and　applied　the　methods　to　solve　practical　emergency　problems.　First,　a　reasonable　decision　index　system　was　constructed　using　latent　Dirichlet　allocation　thematic-cluster　analysis　of　the　focus　word　frequency　from　online　public　opinions.　Python　programming　was　then　used　to　capture　the　big　data　of　social　media　comments　on　emergencies　in　real　time　to　obtain　multisource　comment　data,　which　were　preprocessed　and　visualized.　Second,　based　on　the　optimized　SnowNLP,　emotional　tendencies　and　statistical　analyses　were　performed　on　the　processed　comment　data,　resulting　in　two　quantitative　decision　matrices,　which　were　represented　as　single-valued　neutrosophic　numbers　(SVNNs)　and　probabilistic　linguistic　terms　(PLTs).　Next,　the　bidirectional　projection　method　was　extended　to　an　environment　of　SVNNs　and　PLTs,　and　the　alternatives　were　sorted　and　selected.　Finally,　a　sudden　natural　disaster　event　was　used　as　a　numerical　case　to　verify　the　proposed　method.　Through　a　comparative　analysis,　the　experimental　results　show　the　practicability　and　feasibility　of　the　proposed　decision-making　method,　demonstrating　the　superiority　and　intelligence　of　our　research.　Our　study　can　conduct　real-time　monitoring　of　emergencies　and　intelligently　and　objectively　obtain　quantified　decision-making　data,　thereby　providing　auxiliary　decision-making　support　for　relevant　emergency　management　departments,　ensuring　the　positive　development　of　public　opinion,　social　stability,　and　the　safety　of　people＇s　lives　and　property.　©　The　Author(s)　2025.