EClerk office assistant

Abstract

For decades, people have continued to collect an inordinate amount of paper documents containing important information that should be easily accessible. This paper clutter inhibits indexing this information and easily searching through it. This thesis presents the code architecture and user interface design of the Electronic Clerk, a proof-of-concept electronic office assistant. The Electronic Clerk (EClerk) is a device to assist in reducing paper clutter in the office environment. The device takes paper and speech as input, performs data binding between input streams in order to attach metadata to each document, and structures the data using the Resource Description Framework (RDF) standard. The hardware structure of EClerk consists of a dedicated computer, video camera, scanner, touchscreen, and microphone for capturing input. The software structure consists of the Galaxy speech recognition system, the Haystack information client for retrieval and modification of the collected data, optical character recognition, and a graphical user interface that provides continuous feedback to the user. Primary design principles for this device include providing continuous user feedback and robustness to imperfect input in order to provide a truly usable system.