Decolonizing Knowledge Organization: Indigenous Knowledge Systems in Library Classification

Abstract

Purpose: Library classification systems, like the Dewey Decimal Classification (DDC) and the Library of Congress Classification (LCC), have been underpinned by Western epistemologies, which often marginalize or misrepresent Indigenous Knowledge Systems of management. This system of colonialism denies access to Indigenous knowledge and does not reflect Indigenous worldviews, languages, and relationships with land and community. The objective of this research is to offer an artificial intelligence (AI) facilitated method of decolonizing library classification through the identification and reclassification of Indigenous knowledge materials in a manner that is respectful and representative of Indigenous thought. Methodology: A corpus of more than 500 libraries was compiled. Preprocessing consisted of data cleaning, normalization of subject heading, and removing duplicate and irrelevant records. Feature extraction was term frequency-inverse document frequency (TF-IDF) and word embedding to extract semantic patterns in the metadata and descriptions. This research proposed a new tool, Malleable Migrating Birds Optimized Bidirectional Encoder Representations from Transformers (MMBO-BERT)-based classification tool, which picks culturally insensitive expressions, detects misplaced information, and suggests placement in Indigenous taxonomies and themes determined in consultation with Indigenous scholars. Results: Python was used for the implementation and experimental results demonstrate the research using human expert judgment to classify misclassifications and suggest reclassifications. The MMBO-BERT methodology performed better in Accuracy at (93.1%), Precision at (91.8%), Recall at (94.3%), F1-Score at (93.0%), and Cohen’s Kappa at (0.83). Conclusion: Decolonization of knowledge organization is possible through the use of AI, which can enable libraries to reclassify classification systems more inclusively.