Internet Research Support System: An Application for Immediate Feedback in Authentic Internet Research

Abstract

A prototype Internet Research Support System is presented based on earlier work on the modeling of online reading behavior and evaluating the reading process and its outcomes. Users complete a reading task by browsing the Web and writing a short essay. The online browsing data is captured in real-time, which, along with the essay, reflects the participants’ online reading process. Verbatim Quotient Detection is possible which monitors the proportion of material copied from the Internet. A Latent Semantic Analysis component to evaluate sentence-to-sentence cohesion in the final essay is also described, and is in the process of being integrated with the main system. The system evaluation capabilities provide the participants with immediate feedback that is specific to their final essay in order to help them increase their online reading comprehension. It also archives data for research and evaluation purposes. One hundred and seventeen undergraduate students and one faculty member participated in two experiments to test the system where the outcomes were marked by the system and human markers. Analysis of the online reading activity demonstrated a negative correlation between the human marker’s score and the number of verbatim words in the final essay. Furthermore, the analysis showed that when the whole Internet was available, even with limited time and a common task, there was significant variation in participants’ search strategies, search terms and research processes.