Semantic similarity analysis between the student's reports with the authoritative sources.

Knowledge Building as a pedagogy supports collaborative work between students to improve ideas. The result of knowledge-building discourse in student communities is the development of academic artifacts. Student reports are academic artifacts that can result from a knowledge-building process during their learning. However, the automatic analysis of student reports is quite challenging due to the documents’ length and written language. Given this context, our goal is to analyze the semantic similarity between student reports and curriculum literature using the information coverage measure based on the Skip-gram word2vec model. Word2vec is a well-known set of techniques that provides a way to create a representation of words to predict the nearby words between every word and its context to capture the internal semantic and syntactic information. In this work, through the experimental comparison between the vocabulary list, the best similarity value is 0.90 using word2vec. The results show that each student report was aligned with each document in the curriculum literature using the information coverage measure based on word2vec.