OpenEssayist: An automatic feedback system that supports University students as they write summative essays
OpenEssayist is an automated, interactive feedback system designed to provide an acceptable level of support for students as they write essays for summative assessment. There are two main components to the system: (1) a linguistic analysis engine[dgf1] and (2) a web application that generates feedback for students (Van Labeke et al., 2013[dgf2] ). The main pedagogical challenge in the e-assessment of free text is how to provide meaningful “advice for action” (Whitelock, 2011) in order to support students writing their summative assessments. We have built a first working version of the system in which we use unsupervised graph-based ranking algorithms (following Mihalcea & Tarau, 2005) to automatically extract key words, phrases and sentences from student essays. We have developed several external representations of these summarisation techniques. For example, key words and key phrases can be viewed in a word cloud or in a dispersion graph, and they can be explored and organised into groups. Holistic approaches have also been tested using 'mash-ups' where key words and key sentences are highlighted in context in the essay itself, helping students to investigate the distribution of key words and its potential implications for the clarity of the narrative. This paper will report the findings from our pilot studies of the interactive feedback models associated with the summarisation techniques.
[dgf1]I am not sure 'learning analytics engine' is the best term to use here. Are you deliberately using this, because it connects this paper with your learning analytics work? If it's not important to you, I would put 'linguistic analysis engine' or 'linguistic engine'.
[dgf2]I don't include references in abstracts. If I did include one, it would be to ensure there was no confusion about the kind of work I was referring to, and it wouldn't be a self reference. Personally I would take them all out.