Web-based system that generates adaptive hypertext explanations using a generic user modelling framework.

The goal of our dynamic hypertext generation system, HYLITE+, is to produce encyclopaedia-style explanations of domain terminology. The corpus analysis of online encyclopaedia and previous empirical studies have shown the positive effect of additional information -- e.g., definition of key vocabulary, less-technical content, supply of background information and illustrations -- on the subjects' reading comprehension and reading behaviour. On the other hand, hypertext usability studies have shown that people read 25 per cent slower on the screen, so hypertext needs to be concise with formatting, that facilitates skimming. Our empirical studies have shown that users prefer different additional information depending on the chosen formatting and desired explanation length.

The majority of existing language generation systems have a pipeline architecture which offers efficient sequential execution of modules, but does not allow decisions about text content to be revised in later stages. However, in some cases choosing appropriate content can depend on text length and formatting, which in a pipeline architecture are determined after content planning is completed. Unlike pipelines, interleaved and revisionbased architectures can deal with such dependencies but tend to be more expensive computationally. Since our system needs to generate acceptable hypertext explanations reliably and quickly, the pipeline architecture was modified instead to allow additional content to be requested in later stages of the generation process if necessary. in such an architecture it is not possible to take into account text formatting and length (which are determined towards the end) when choosing the text content (which happens in the beginning).

Architecture