A study of the description of alt text suitable for different types of posts

Abstract

Due to the rapid development of the Internet and mobile devices social software has increased Social objects on social networks are becoming diversified and images and videos are gradually replacing text as the primary social object The visually impaired are also constant users of social software There are currently accessible reading tools that can aid them in using social software However there are still some dilemmas that accessible reading tools cannot currently solve; for example images do not have alt text To improve the visually impaired user experience Facebook has launched an automatic alt text system in recent years This system combines artificial intelligence that recognizes images and alt text syntax which can automatically generate alt text for images The descriptions currently produced by the system are still very monotonous and there is always room for improvement Therefore this study reviews another service called Audio Description that can provide visual information for the visually impaired Generally objectivity is the top purpose for Audio Describer descriptions In recent years some scholars have suggested that Audio Describer should produce Audio Descriptions more dramatically and subjectively to improve user experience However the suitable narrative method for Audio Description should not be based on a subjective or objective dichotomy It should be adjusted according to the type of video content in order to produce a better user experience Therefore this study takes the literature related to Audio Description as a reference and proposes that the alt text of Facebook posts should be adjusted according to the types of those posts In this study the Conceptual Model is used to build three types of alt texts of twenty posts with ten kinds of news content and two types of graphic and textual relationships The study examines the relationship between graphics and text that is suitable for producing generic specific or abstract alt text in a variety of content areas The study found that because successive posts spend a smaller percentage of time explaining image information in the text content these posts are suitable for specific alt text The text of illustrated posts contains a lot of image information so these posts are appropriate for specific or abstract alt text Finally the aggregation of the research results organized into a flow chart of the different ways of producing alt text provides a reference for future related research or the development of related services

Date of Award	2020
Original language	English
Supervisor	Chun-Heng Ho (Supervisor)