Our system automatically generates tailored ad text based on the travel magazine advertisements layout and multiple travel blogs. The user inputs the layout of the travel magazine advertisement, which is then processed by a prediction-evaluation model (PEM). This model predicts the required number of text words and font size in the layout based on the input layout. Next, feed the data generated by the PEM and multiple tourism blogs into the text generation model, and ultimately produce text that meets the requirements and fits the designated text area in tourism magazines' advertisements. The generated text is then placed in the text area of the layout. It includes our new self-attention layer to ensure that the output text contains essential information. We conducted extensive experiments on real datasets to demonstrate the effectiveness of our proposed model. Using our model(Adaptive-Bert), users can create travel magazine advertising text that meets their requirements. Our method ensures that the generated text is of high quality and meets layout requirements.