Machine Learning For Classifying Bibliographic Resources: Using topical headings to infer alignments to Australian Curriculum learning areas

Since 2017 the Schools Catalogue Information Service (SCIS) has enabled users to search the SCIS catalogue by curriculum learning area. A rules-based algorithm is used to infer the learning area to which records are aligned. This paper explores the possibility of using machine learning to supplement or replace the current approach. Multi-label supervised classifiers were trained on topical headings from a large dataset of digital learning resources from the Scootle repository. They were then tested on a smaller set of SCIS records and demonstrated adequate results for a subset of learning areas, with better precision than recall. Methods for improving classification are discussed.