Customizable Information Systems
Caroline M. Eastman
In this research area we consider several aspects of customizability in information systems. Such systems can be modified either by the user (adaptability) or by the system (adaptivity) to accommodate user preferences or needs or system goals. There is a wide range of potential customizable modifications, including simple selection of colors, conversion from text to speech, and maintenance of user histories or profiles. Customization capabilities can be used to create personalized systems. Clearly some level of possible customization is desirable to accommodate physical and sensory limitations and differences in language and custom. However, providing customization options involves a tradeoff between the amount of customization and the resources required. Our previous work in this area includes the development of a model of customization and some investigations of customization options and tradeoffs.
We have developed a model (Customizable Information Components) to describe the structure of customizable information systems that are able to present information using different formats and media (Eastman and Kim, 1999); this model was developed based upon preliminary work done by Su Hee Kim in her doctoral dissertation in computer science. The primary components of this model are BICs (base information components), DICs (derived information components), and AICs (associated information components). BICs represent the original information. A DIC is derived from a CIC by applying a transformation or transcoding that may or may not be reversible. An AIC contains additional information about the BIC but is not derived from it. Consider examples from document retrieval. The documents are BICs. Possible derived components include keyword lists and translations. A review of the document is an AIC; it contains information relevant to the document BIC but cannot be derived directly from it.
Two specific projects have investigated customization tradeoffs. Work with Douglas Hutchison (MS graduate in Computer Science) and Terry Tirrito (College of Social Work) developed a customizable user interface and system for a vintage movie trivia game (Hutchison, Eastman, & Tirrito, 1997). The design of the system was based upon a survey of 122 older adults. It included the ability to change font and button size and included easy to use variations of list boxes and combo boxes. Another project with Su Hee Kim (PhD graduate in Computer Science) investigated general issues in information presentation and transformation in hypermedia systems. One of the specific issues addressed was the question of hypermedia node size (Kim & Eastman, 1999). Two different window sizes and two different node lengths were compared. The window size did not have a significant impact. However, the longer nodes (which were based upon logical breaks in the text) were more effective than the shorter ones (which were based upon the window size). Effectiveness was evaluated by asking subjects to complete a series of fact retrieval tasks and measuring the time required and the number completed successfully.
Another more recent project examined the potential impact of personalized product presentations in an e-commerce environment; this work was conducted with Nancy Lightner of the Moore School of Business (Lightner & Eastman, 2002). People differ in their dominant information processing style; some are predominantly verbal, and others visual. An experiment using a simulated e-commerce site compared sentential (verbal), diagrammatic (graphical/visual), and mixed (both sentential and diagrammatic) presentations. A survey was used to assess user satisfaction. Both verbal and visual users preferred the sentential web site to the diagrammatic web site. Although many users expressed a desire for both words and pictures, the satisfaction measured for the mixed site was not significantly greater than for the sentential site. These results indicate that customization based upon information processing style may not be cost-effective for an e-commerce site.
We plan further work investigating customization tradeoffs, especially in the area of e-commerce. These tradeoffs involve consideration of user preferences, user behavior, technical feasibility, resource use, and platform dependence. An important issue is that of the granularity of customization; if customization is supported, how many options should be given? Another important issue is the relationship between user preference and effectiveness (either from the perspective of the user or the perspective of the system owner).
Eastman, C. M. and Kim, S. H. (1999) Customizable information components. Customized Information Delivery Workshop. A workshop presented in conjunction with the 22nd Annual International ACM SIGIR Conference on Information Retrieval. Berkeley, CA, August 19, 1999, pp. 1-5. http://www.ted.cmis.csiro.au/sigir99/
Hutchison, D. Eastman, C. M. & Tirrito, T. (1997) Designing user interfaces for older adults. Educational Gerontology, 23, 497-513.
Kim, S. H. & Eastman, C. M. (1999) An experiment on node size in a hypermedia system. Journal of the American Society for Information Science, 50(6), 530-536.
Lightner, N. J. and Eastman, C. M. User preference for product information in remote purchase environments. Journal of Electronic Commerce Research, 3(3), 174-186.