Developing an Original Computational Framework for Indigenous Knowledge Representation in Nigerian Agricultural Data Systems

Indigenous agricultural knowledge systems in Nigeria encode centuries of locally adapted practices for crop management, weather prediction, and pest control that formal data systems fail to represent, and developing an original computational framework for their capture and integration represents a significant interdisciplinary contribution. This study developed an original computational framework for representing and integrating indigenous agricultural knowledge in Nigerian agricultural data systems. A multi-phase methodology was employed: systematic review of indigenous knowledge representation approaches (52 publications from 2018 to 2024), ethnographic knowledge elicitation with 120 farmers from six geopolitical zones using structured knowledge capture protocols, ontology development through expert consultations with 20 agricultural scientists and 8 knowledge representation specialists, and technical framework design and validation. The ethnographic work documented 847 distinct indigenous knowledge statements covering planting calendars, intercropping arrangements, natural pest management, and soil fertility indicators from six indigenous agricultural systems. The computational framework proposes an Agricultural Indigenous Knowledge Ontology (AIKO) with five concept categories: ecological indicators, management practices, temporal rules, social knowledge transmission, and validation evidence. A knowledge integration protocol specifying how AIKO-encoded knowledge can augment machine learning models for crop recommendation and yield prediction is developed and validated on a cassava recommendation task in Anambra State, showing 14.7 percent accuracy improvement over the model without indigenous knowledge integration. Expert review confirmed the framework's original contribution to knowledge representation and data science for African agricultural contexts.

Keywords: indigenous knowledge, agricultural data systems, ontology, Nigeria, knowledge representation