International Journal of Computer Science and Technology

International Journal of Computer Science and Technology | Vol. 9, No. 2, February 2007 | pp. 9–16

DOI: 10.46882/2007/IJCST/090230

Research Article

Title: Semantic Data Schema Alignment Across Heterogeneous Web Repositories Using Neural Embeddings

Names of Authors: Min-Seok Kim¹, Aparna Rao²

Authors’ Affiliations: ¹Department of Computer Science and Engineering, Seoul National University, Seoul, South Korea; ²Department of Computer Science and Automation, Indian Institute of Science, Bangalore, India

Abstract:
Integrating enterprise information systems often requires mapping structural schemas that use different naming metrics and metadata design standards. This paper proposes an automated schema alignment framework that processes element semantics using multi-dimensional neural context embeddings. The matching engine evaluates lexical attributes alongside hierarchical dependency constraints via a modified graph similarity algorithm. The alignment framework was evaluated on an e-commerce database benchmark containing 80 distinct relational structures. The model achieved a matching precision rate of P = 92.15% and an overall F1-measure index of 89.40%, outperforming traditional text-string edit-distance matchers by 16.4%. The computational complexity scaled linearly at O(N log N) relative to node density, providing an efficient approach for real-time schema mapping within large enterprise information warehouses.

Keywords: Schema Alignment, Semantic Web, Data Integration, Neural Embeddings, Graph Similarity

Manuscript Timeline: Received November 05, 2006; Revised December 14, 2006; Accepted January 10, 2007; Published February 23, 2007

Citation: Kim, M. S., & Rao, A. (2007). Semantic Data Schema Alignment Across Heterogeneous Web Repositories Using Neural Embeddings. International Journal of Computer Science and Technology, 9(2), 9–16.

Subscribe to read the full article: https://www.internationalscholarsjournals.org/subscribe-to-read