Reflective Writing Analytics (RWA) is the computational analysis of reflective writing. That is when an author writes about themselves, their experiences, and their feelings and thoughts about these topics.
The usage of Reflective Writing Analytics as a term emerged informally during the LAK15 conference (March 2015 in Poughkeepsie, NY) with the initial documentation of the term in a blog by Simon Buckingham Shum, and more formally in Journal of Learning Analytics Paper: Towards the Discovery of Learner Metacognition From Reflective Writing (authored by myself, Kirsty Kitto, and Peter Bruza).
A requirement for the computational analysis of reflective writing, is a productive way of working between the psychosocial epistemic domain of reflective writing and the computational epistemic domain of analytics. Computation is not necessary for the analysis of reflective writing; it is possible to conceive of a purely psychosocial process of reflective writing analysis. Similarly, it is possible to conceive of computational analysis of textual data without any psychosocial interaction. However, RWA involves an interdependence between the psychosocial and the computational; that typically the flow of information is initiated in the psychosocial domain, is translated into the computational where it undergoes some level of computational processing, and is then translated back into the psychosocial domain where it obtains meaning appropriate to the context in which it originated.
In my doctoral thesis, I proposed a conceptual model of RWA based on eight aspects, where each aspect (edge) represents the interaction between two activities (nodes). Three of these aspects describe the psychosocial epistemic domain, three describe the computational domain, and the other two are transepistemic, that is they bridge the two epistemic domains. A brief description of the eight aspects is provided below:
- Interpreting–Reflecting is the person’s construction of meaning through interpreting situations in relation to self, and reflecting on that meaning and its implications for action.
- Reflecting–Writing involves the person expressing meaning through their own story. It is a personal narrative about selected ideas that arise from self-reflection. It results in the artefact of reflective writing.
- Writing–Interpreting provides psychosocial explanations based on the interrelationship between the written expression and the person’s own interpretation of that expression. It makes explicit that any explanation requires both the person’s story and their personal interpretation of it.
- Symbolising–Processing is the construction of computational representations of the artefact resulting from the person’s writing. It involves representing features of the reflective text, from simple direct representation of lexical features through to complex modelling of psychosocial phenomena.
- Processing–Analogising involves the expression of the machine generated analytics in a form suitable for human interpretation. It requires both the anticipation of usefulness of the resultant analytics, as well as appropriate computational processing of the information to create them.
- Symbolising–Analogising provides computational explanations based on the interrelationship between the original representations of the reflective text and the analytics prepared for interpretation. It makes explicit that explanation requires the output ‘Analogising’ of computation to be understood in terms of the input ‘Symbolising’ of the text.
- Writing–Symbolising facilitates the transfer of psychosocial representations to computational representations. This aspect involves the translation of word symbols into computational symbols, and in doing so aims to take psychosocial characteristics and model them computationally. This aspect provides `Representational Translation’ between the epistemic domains.
- Analogising–Interpreting facilitates the semantic transfer from the computational to the psychosocial. It is concerned with meaning-making through the psychosocial interpretation of the computational analytics. This aspect provides `Semantic Translation’ between the epistemic domains.
You can find more information on this conceptual model of RWA in my thesis located on my QUT ePrints Page.
Much of my recent work on RWA is referenced on the UTS CIC A3R Project Page and the LAK’17 Conference Paper about the work: Reflective Writing Analytics for Actionable Feedback.