Updating database schemas without breaking the UI: modeling using cognitive semantic categories

Abstract

Data management user interfaces are ubiquitous in information systems and web-based applications. From the oldest spreadsheet to the most modern database, end users and administrators alike have interacted with tabular data. Usually, each concept is represented by a table and columns. Change to the structure of each concept requires structural change to the tables and columns, which is costly. Tailor-made database and web applications may overcome this obstacle by designing UIs on top of the data layer, providing some degree of data independence. However, changes in their schemas do not automatically propagate into the user interface, and so their maintenance is expensive. In this paper we present a user interface that lets the end user alter the schema without the need for programming skills, eliminating the need for expensive software maintenance. To this end we propose an automatically generated user interface to include schema and data management functions. We built and evaluated an Adaptive Information System user interface (AIS UI), incorporating schema evolution functionality. In usability testing, first-time users were able to perform various data management tasks equally fast or faster than users using Microsoft Access, and on average ~43% faster than users using Microsoft Excel. Task completion rates using the AIS significantly exceeded those using Microsoft Access and were comparable (>95%) with those using Microsoft Excel.