As a medium of high-frequency access, websites highly affect our life. Due to the specific and subjective content selection criteria of viewers, they prefer to click on pages that they think look comfortable to them. Some viewers prefer better UI design whereas others prefer ease of access where they do not have to spend much time on the website to find what they are looking for. Therefore, designing web pages which attracts the attention of Internet users helps increasing the number of views which in turn results in increased economic benefits. A reasonable aesthetic assessment of a web design has become an important research topic. Since the aesthetic assessment of web design has great subjectivity, it becomes very difficult to correctly evaluate the web page aesthetics. For web page assessment, this paper divides the task into multiple modalities. In this paper, multiple modes are evaluated separately, and a more accurate assessment of each mode is presented. Assessments of the various modalities are combined to get an overall eigen vector. A complex and subjective assessment task is decomposed into multimodal subtasks and a more specific and accurate feature vector is obtained by merging the subtask assessments. This method can solve the complex problem of subjective assessment tasks which, otherwise, are difficult to solve directly through training model. By conducting several subjective experiments and comparing the outcome with the results predicted by the model, it is shown that the proposed model is correlated well to the subjective measurement of the viewer and provide a strong reference for the aesthetic assessment of a web page.