Before students self-validate their work, they upload a writing sample. The writing sample is a piece of the student's own previous writing. Validate compares later submissions against it to assess Authorship Continuity.
Authorship Continuity asks one question: does this submission read like it came from the same writer as the student's writing sample?
There are three results, consistent, inconclusive and divergent.
A change of genre can make two pieces by the same student read differently. Studiosity's Validate handles this in several ways:
It's built for academic writing. The model is designed for undergraduate and postgraduate academic writing, including essays, scientific and lab reports, and reflective writing.
It leans towards the student when results are unclear. If the model can't be confident, it returns Inconclusive rather than a result either way. It's also tuned to recognise the same writer, so fewer students are wrongly shown as Divergent (writing that differs noticeably from the sample).
A Divergent result isn't a finding of misconduct. A different type of assignment is one of the expected, legitimate reasons writing can differ. So are a different style or an older writing sample.
It's one of three signals. Human Factors and the student's answers to the questions are weighed alongside it. A student whose writing doesn't match the sample can still be validated on the other signals.