This　research　proposes　a　method　for　analyzing　student　social　relationships　based　on　three-dimensional　spatio-temporal　co-occurrence.　By　leveraging　detailed　timestamp　information　from　student　behavior　data,　the　method　constructs　a　three-dimensional　co-occurrence　tensor　encompassing　location,　date,　and　time　dimensions.　Two　key　features,　three-dimensional　co-occurrence　spatiotemporal　diversity　and　weighted　co-occurrence　frequency,　are　extracted　from　this　tensor.　A　multiple　linear　regression　algorithm　is　then　employed　to　calculate　the　social　intensity　values　among　students.　Additionally,　the　Louvain　community　detection　algorithm　is　utilized　to　mine　hierarchical　community　social　relationships　from　the　constructed　student　social　network.　©　2024　Copyright　held　by　the　owner/author(s).