MACSc Project Presentation - Constructing training data for large-scale research topic classification
Who: Xiaoke Xu
Title: Constructing training data for large-scale research topic classification
Evangelos Milios - Faculty of Computer Science (Supervisor)
Fernando Paulovich - Faculty of Computer Science (Reader)
The objective of this research is to construct training data for the hierarchical classification of research papers in Natural Sciences and Engineering into NSERC Evaluation Groups (corresponding to disciplines) and Research Topics (corresponding to research areas within a discipline). This project builds on previous work on this problem, in which training data was manually selected and limited to three evaluation groups. Our work is efficiently scaling up to all NSERC twelve evaluation groups and their research topics automatically, with only minor user involvement through an interactive visualization to remove outliers based on the 2D projection of training data.
Room 311, Goldberg Computer Science BuildingMACSc Project Presentation - Constructing training data for large-scale research topic classification