Improving SDG Classification with Topic Models and Combinatorial Fusion
Abstract
Combinatorial fusion analysis (CFA) is a machine learning and artificial intelligence (ML/AI) paradigm for combining multiple scoring systems using the rank-score characteristic (RSC) function and cognitive diversity (CD). In this work, CFA is used to combine two topic models A and B to improve the classification precision. Each of these two models measures how similar the contents of a publication (or a document) are to each of the 17 Sustainable Development Goals (SDGs) of the United Nations. Each of the individual models is characterized and analyzed using the RSC function. The classification results of models A and B and combined models from score and rank combinations are also evaluated. Classification precision is calculated using Pre@k, k = 1, 3, 5, and 8 when compared to classification results obtained by human experts.
Subject Area
Computer science|Information science
Recommended Citation
Orazbek, Ilyas, "Improving SDG Classification with Topic Models and Combinatorial Fusion" (2021). ETD Collection for Fordham University. AAI28718454.
https://research.library.fordham.edu/dissertations/AAI28718454
