To　further　explore　the　impact　and　hierarchical　transfer　strategies　of　different　passengers　on　public　transportation　(PT)　under　the　condition　of　major　epidemic,　an　individual　PT　travel　chain　was　extracted　and　travel　knowledge　graphs　were　constructed　based　on　the　multi-source　PT　trip　data.　The　K-means　algorithm　was　used　to　identify　passengers＇　PT　dependence　hierarchies.　Then　the　improved　Apriori　algorithm　was　used　to　mine　the　strong　association　rules　with　different　itemset　length,　and　the　dependence　level　transfer　incentive　policies　were　proposed.　The　results　show　that　there　is　a　negative　correlation　between　the　number　of　antecedents　of　the　strong　association　rules　and　their　support　degree　under　major　epidemic　conditions.　The　lower　the　PT　dependence　levels,　the　lower　the　co-occurrence　degree　and　occurrence　probability　of　association　rules　indicators.　The　total　distance　from　home　and　destination　to　the　PT　stations,　car　availability　and　the　support　degree　of　relatives　and　friends　to　use　PT　are　the　key　indicators　for　each　PT　dependence　group　to　be　improved,　while　whether　having　the　routes　in　high-risk　areas　during　the　epidemic　period　is　the　most　key　improvement　indicator　of　the　PT　usage　behavior　for　the　relatively　dependence　groups.　The　study　is　conducive　to　the　formulation　of　PT　policy　and　the　promotion　of　PT　service　during　major　and　post-epidemic　period,　and　provides　support　for　ensuring　and　balancing　the　proportion　of　green　travel　structure　in　urban　travel.　©　2022,　Editorial　Department　of　Journal　of　Southeast　University.　All　right　reserved.