Multi-attribute group decision making (MAGDM) model which uncovers the preferences of individuals has become increasingly important in enhancing the efficiency of expert systems. Meanwhile, the use of flexible linguistic expression (FLE) and the personalized individual semantics (PIS) bring challenges to MAGDM. Therefore, in this paper, we devote to design a method based on personalized individual semantics for MAGDM with flexible linguistic expression, named PIS-based FLE-MAGDM. In the PIS-based FLE-MAGDM, we firstly develop a consistency-driven PIS model to derive personalized numerical scales (PNS) for linguistic terms with flexible linguistic preference relation (FLPR) and obtain the heterogeneous attribute weight by different decision makers. Then, we propose the individual selection process for getting the individual ranking of alternatives. Next, we provide the collective selection process to aggregate the individual ranking into the collective ranking. Finally, an application under the background of airline selection and some comparative analyses are presented to illustrate and justify the effectiveness of our proposals.