test
Search publications, data, projects and authors

Free full text available

Article

French

ID: <

10670/1.fnmgle

>

Where these data come from
Building knowledge by exposure to a course with LSA

Abstract

National audience The learning paradox, i.e. the possibility to derive new knowledge from existing knowledge, is a problem that has been addressed by many researchers. Landauer and Dumais (1997) propose a computer application, latent semantic analysis (LSA, for Latent Semantic Analysis), a multidimensional factor analysis method for analysing, on the basis of co-occurrences of words, semantic correlations of words or paragraphs. In particular, this model has been used to simulate the acquisition of human vocabulary, and we are replying here to this work on exposure to a course. A first test allows us to examine the evolution of the semantic interword linkages, calculated by LSA, as exposure to this course takes place. Three types of change in proximity can be identified, leading to its being considered similar to that of humans. A second test compares the scores of students to a multiple-choice questionnaire with those of LSA, once he has processed the course given to the students. LSA makes it possible, without any prior human coding, to report on the building of knowledge by exposure to a course.

Your Feedback

Please give us your feedback and help us make GoTriple better.
Fill in our satisfaction questionnaire and tell us what you like about GoTriple!