Enhancing Query Expansion with ML-SSO: A Machine Learning, Semantic, Statistical, and Ontological Framework

This paper presents a Machine Learning -Semantic Statistical and Ontological query expansion approach to enhance the performance of information retrieval systems. The proposed methodology ML-SSO looks to combine the strengths of each technique to obtain relevancy and accurately expanded queries. In this, pre-trained word-embed models were used for contextually relevant term expansions, and domain-specific ontologies were used to provide accurate and custom query expansions. Large-scale experiments conducted on benchmark datasets demonstrated that the ML-SSO outperformed traditional ones in MAP, Precision at k (P@10), Recall, and F-measure scores. These results make clear that the ML-SSO query expansion method improves the relevance and accuracy of the Retrieved Document. Future work will be directed towards optimizing the scalability, incorporating dynamic ontology driven into it, and using other advanced models in machine learning to further increase the performance of the system.