Given the scarcity of structured linguistic resources in Assamese, we propose a rule-based relation extraction system that identifies semantic relationships between pairs of entities within sentences. Raw texts were sourced from Assamese Wikipedia biographies, and entities were categorized into seven semantic types. A set of linguistically informed rules was developed to extract 10 predefined relation types between entity pairs. The system was evaluated on a manually curated dataset of 500 sentences, achieving high precision 71.2% and an F1-score of 76.3%, demonstrating the effectiveness of rule-based methods in this low-resource language. This work contributes a new annotated dataset and forms a foundation for future semantic applications such as knowledge graph construction in Assamese.