圖片描述技術輔助文本於可讀性模型準確度之研究

“Adaptive reading” is the process of adapting the difficulty level of a text to the ability of the reader so that the reader can understand the text more easily. However, with the large number of books being published, it is time-consuming and labor-intensive to manually assess the difficulty level, and therefore, efforts have been made to develop readability models. In the past, the development of text readability models was often characterized by text, and the content of pictures was not considered as a factor in assessing the difficulty of text. However, it has been suggested in other fields that pictures can affect reading comprehension. Inspired by this research, this study will utilize picture description techniques to extract the content of pictures as features to be added to the text readability model, in order to investigate whether the addition of picture content can improve the accuracy of the readability model.