To　solve　the　problem　of　strong　subjectivity　and　low　credibility　of　complaints,　a　new　detection　model　of　false　information　in　environment　complaints　based　on　graph　convolution　was　proposed.　Firstly,　the　text　of　environment　complaint　is　preprocessed:　then　the　word　in　the　complaint　text　and　lexicon　are　used　as　nodes,　the　edges　between　words　are　constructed　by　point　mutual　information,　the　edges　between　text　and　words　are　constructed　by　term　frequency-inverse　document　frequency,　and　the　edges　between　text　and　text　are　constructed　by　text　similarity.　The　edges　between　words,　text-word　relationship　and　text-text　relationship　are　used　to　construct　the　text　graph　of　environmental　complaint　based　on　word　co-occurrence,　text-word　relationship　and　text-text　relationship;　finally,　the　node　information　is　passed　between　nodes　through　graph　convolution,　and　the　coupling　between　complaint　information　is　used　to　detect　complaint　false　information.　Through　designing　experiments,　the　model　is　compared　with　the　methods　of　common　classification　models,　and　the　experimental　results　show　that　the　model　has　good　performance.　It　is　suitable　to　be　applied　in　the　task　of　environment-based　false　complaint　detection.　©　2023　SPIE.