Automated free text marking with Paperless School

The Paperless School automarking system utilises a number of novel approaches to address the challenge of providing both summative and formative assessments with little or no human intervention. The Paperless School system is designed primarily for day-to-day, low stakes testing of essay and short-text student inputs. It intentionally sacrifices some degree of accuracy to achieve ease of set up, but nevertheless provides an accurate view of the abilities of each student by averaging marks over a number of essays. The system is designed to function as a back-end service to an Learning Management System (LMS), thus facilitating the marking of large numbers of texts. This should enable considerable teacher resources to be freed up for other teaching tasks. In this paper we will discuss some of the issues involved in bringing computational linguistics to bear in the educational context. We will cover • how Blooms Taxonomy (the pedagogical model underlying most formal grading schemes) can be represented in software. • an overview of the steps required to derive a grade that will sufficiently closely predict the grade a human marker would give. • extending the system to include formative assessment, via intelligent comment banks.