SignDIn: Designing and assessing a generisable mobile interface for Sign support

Abstract

SignSupport is a collaborative project between the Computer Science departments of the University of Cape Town (UCT) and the University of the Western Cape (UWC), South Africa. The intention of the software is to assist Deaf users to communicate with those who can hear in domain-specific scenarios.The penultimate version of this software is a mobile application that facilitates communication between Deaf patients and hearing pharmacists through the use of Sign Language videos stored locally on the mobile device. In this iteration, adding any new content to the mobile application necessitates redevelopment, and this is seen as a limitation to Sign Support. The architecture hinders the addition of new domains of use as well as extending the existing domains. This Dissertation presents the development and assessment of a new mobile application and data structure, together called SignDIn, and named as an amalgamation of the words 'Sign', Display' and'Input'. The data structure facilitates the easy transfer of information to the mobile application in such a way as to extend its use to new domains. The mobile application parses the data structure, and presents the information held therein to the user. In this development, the Dissertation sets out to address the following:1.How to develop a generalisable data structure that can be used in multiple contexts of Sign Language use.2. How to test and evaluate the resulting application to ensure that parsing the data structure does not hinder performance. The first objective of this research aims to develop a data structure in a generalised format so that itis applicable to multiple domains of use. Firstly, data structure technologies were evaluated and XML selected as the most appropriate out of three candidates (Relational Databases and JSON being the other two) with which to build the data structure. Information was collected from the International Computer Driver's Licence (ICDL) and Pharmacy domains and an XML data structure designed passing through three stages of development. The final outcome of the data structure comprises two XML types: display XMLs holding the display information in a general format of screen, video, image, capture and input; and input XMLs holding the list of input options available to users. The second objective is to test the performance of the mobile application to ensure that parsing the XML does not slow it down. Three XML parsers were evaluated, SAX Parsing, DOM Parsing, and the XML Pull Parser. These were evaluated against the time taken to parse a screen object as well as the screen object throughput per second. The XML Pull Parser is found to be the most efficient and appropriate for use in SignDin.