No abstract

Form-intensive Web applications are common among institutions that collect bulks of data in a piecemeal fashion. European funding programs or income tax return illustrate these scenarios. Very often, most of this data is already digitalized in terms of documents, spreadsheets or databases. The task of manually filling Web forms out of these resources is not only cumbersome but also prone to typos. It does not benefit from the fact that the data is already in electronic format. Alternatively, externally-fed autofilling scripts can be programmed (e.g. using iMacros and Visual Basic) to code once, and enact many times. Unfortunately, this approach is programming intensive and fragile upon upgrades in either the website or the structure of the external source. This moves these tools away from users with scarce programming skills. We strive to empower these users by abstracting the way feeding solutions are realised. Since external sources tend to be structured, they offer the chance to be abstracted in terms of models. Autofilling scripts can then be generated as weavings between the external data model and the website model. We describe WebFeeder, a plugin for iMacros that introduces autofilling-script models as first-class artifacts in iMacros. The synthesis, enactment and maintenance of these script models are handled without leaving iMacros, minimizing users' cognitive load and involvement.