Identifying essential elements in justifications of student drafts

Abstract

The proposal draft is the first step to achieve a degree by students in many educational institutions. This proposal is transformed into a thesis after several revisions by an academic adviser. In addition, each proposal must comply with requirements of institutional guidelines. In this paper, we explore a learning approach to identify essential elements: importance, necessity, convenience, and benefits; that are expected to appear in a justification section of a proposal or thesis draft. We present a method based on a Language Model approach. Preliminary results show that the elements of necessity, importance, and benefits obtained acceptable results, considering that this task is complex for an academic adviser. The identification of convenience requires further improvement. A language model based on n-grams showed more consistent and better results than a model based on neural networks. Part of speech tagging contributes to improve results in both language model techniques.