A　key　issue　in　the　management　of　water　resources　in　China　is　the　adoption　of　a　water　supply　total　quantity　control　framework　at　the　regional　level　based　on　socioeconomic　development.　This　necessitates　more　effective　management　of　water　resource　demand　to　achieve　a　balance　between　supply　and　demand.　This　paper　aims　to　conduct　an　in-depth　analysis　of　the　industrial　water　usage　structure,　identify　the　primary　influencing　factors,　and　thereby　provide　insights　for　the　design　of　demand　management　policies.　Using　a　typical　application　scenario　of　＇water　extraction　and　supply＇in　a　subsidiary　of　a　water　supply　company　in　Beijing,　a　water　knowledge　graph　incorporating　multiple　data　sources　is　constructed.　The　study　proposes　a　knowledge　graph　embedding　model　based　on　graph　convolutional　neural　networks.　Over　the　36　months　from　2019　to　2021,　the　water　units　in　the　accommodation,　office,　urban　resident,　healthcare,　retail,　education,　catering,　and　residential　services　industries　are　clustered,　and　the　water　usage　is　analyzed.　The　results　of　this　study　are　expected　to　provide　references　for　optimizing　urban　water　resource　utilization.　©　2023　IEEE.