The development of general-purpose artificial intelligence that can understand and reason with common sense is a significant challenge. However, there has been a lack of research on fully considering personality traits in common sense understanding and reasoning tasks. To address this, we create a personalized commonsense knowledge comprehension and reasoning dataset. This dataset organizes reasoning knowledge with typed if-then relations and variables, while also introducing personality traits as constraints. We adopt a two-stage training framework based on curriculum-learning to gradually improve the model’s personalized commonsense knowledge comprehension and reasoning ability. Additionally, We compare pre-trained language models such as BERT, GPT2, and BART with different structures. The experimental results show that the models trained using the curriculum-learning training framework are able to generate more diversified and personality-trait-compliant commonsense reasoning results.