Carnegie Mellon University
file.pdf (108.8 kB)

Learning to Navigate Web Forms

Download (108.8 kB)
journal contribution
posted on 2004-01-01, 00:00 authored by Anthony Tomasic, William Cohen, Susan Fussell, John Zimmerman, Marina Kobayashi, Einat Minkov, Nathan Halstead, Ravi Mosur, Jason Hum
Given a particular update request to a WWW system, users are faced with the navigation problem of finding the correct form to accomplish the update request. In a large system, such as SAP with about 10,000 relations for the standard installation, users are faced with a sea of thousands of forms to navigate. For familiar tasks, users have various aids, such as personal tool bars, but for more complex tasks, users are forced to search or navigate for the correct form, or forward the update request to a specialist with the expertise to handle the request. In this later case, the execution of the request may be delayed since the specialist may be unavailable, or have other priorities. Also, typically the user and specialist engaged in a time consuming clarification dialog to extract additional information required to complete the request. In this paper we study the problem of building an assistant for the navigation problem for web forms. This assistant can be deployed either directly to a user, or to specialist that receives a stream of requests from users. In the former case the assistant helps the user navigate to the right form. In the latter case, the assistant cuts ambiguous communication between the user and specialist. We present experimental results from behavioral experiments and machine learning that demonstrate the usefulness of our assistant.




Usage metrics


    Ref. manager