[Objective] A novel computational framework for trust in online short-rent platform is proposed. Multiple groups of low-dimension feature subsets are provided to users for presenting personal information, which contributes to ease the problem in trust computations caused by missing data. [Methods] Evolutionary algorithm based rough-set feature selection is used. Image and text analysis are used for feature extraction. [Results] The proposed framework reduces dimension into 5% of the original feature set with classification accuracy remaining unchanged. [Limitations] Further analysis could be done on overseas platform data. [Conclusions] The proposed framework could provide multiple groups of low-dimension feature subsets, which could help to ease the problem in trust computing caused by missing data and privacy.