Home / News / Can a computer be trained to recognise a reflective pharmacy student?

Can a computer be trained to recognise a reflective pharmacy student?

CIC researchers Ming Liu and Simon Buckingham Shum have continued their longstanding collaboration with Cherie Lucas (UTS School of Pharmacy) and a new collaborator, Efi Mantzourani (U. Cardiff). The team has already accomplished a world first for UTS — providing instant feedback to students on their reflective writing. Building on their grammar-based approach, they have now reported the first steps towards training their text analysis infrastructure to classify student writing purely by analysing examples.

In this paper, several statistical machine learning classifiers were trained on a corpus of 301 reflective statements, using emotional, cognitive and linguistic features from the Linguistic Inquiry and Word Count (LIWC) and reflective rhetorical features from Acawriter developed by CIC. The study results showed that a Random-forest classifier performed well on distinguishing between good and poor reflective writing. 

This work was presented at the premier international conference on Artificial Intelligence in Education — read more for the slides or the full paper details:

Liu, M., Buckingham Shum, S., Mantzourani, E. and Lucas, C. (2019). Evaluating Machine Learning Approaches to Classify Pharmacy Students’ Reflective Statements. Proceedings AIED2019: 20th International Conference on Artificial Intelligence in Education, June 25th – 29th 2019, Chicago, USA. Lecture Notes in Computer Science & Artificial Intelligence: Springer. 

Recently, Beijing Consensus on Artificial Intelligence and Education published by UNESCO provides guidance and recommendations on how best to use AI technologies for completing the education 2030 agenda. Find more to access UNESCO official website.

Top