Ontology, a branch of philosophy, studies the nature of reality and the classification of entities. In data science, ontology structures and organizes knowledge, facilitating integration between diverse data sources. Ontology matching is a crucial step in data integration, enabling unified communication across systems and enhancing compatibility, analysis, and decision-making. However, effective integration requires careful planning and methodology selection. This paper proposes a semi-automated approach to data integration based on ontology creation and matching. The creation phase involves manually converting the non-ontology data source into an ontology data source to ensure high accuracy, as it is created by domain experts and contributes to more accurate and effective results in the next phase. In the second phase, the proposed method leverages word embedding techniques, specifically Global Vector (GloVe), which represents words as numerical vectors based on contextual similarity. Cosine similarity measures vector closeness to improve semantic alignment. Experimental results demonstrate the effectiveness and potential of the proposed methodology for improving ontology matching and data integration.