Automated Database Schema Matching Engine

Database Schema Matching is a process which intakes multiple schema as an entry and yields back a mapping that classifies a similar component in these schemas. This process is mostly used to locate and identify semantically related-target. With this method or process it eases the finding and matching of divergent and randomly scattered data sets. It is one valuable tool for data processing and schema integration. Researches shows that various methods for schema matching based on different schema level matcher and classification criteria are proposed in order to find the most similar attributes and element in the schemas. Schema matching is classified into two approaches; Individual Match Approach and Combining Matchers Approach. In this thesis the individual match approach is used, which considers the schema level that is linguistic based. Past studies exhibited several methodologies to make the matching process in schema matching partially and fully automated, while in this thesis Convolutional Neural Network (CNN) methodology is proposed to implement an automated database schema matching engine with the aid of cosine similarity algorithm and Jaro Winkler algorithm. One of the powerful characteristics of the proposed methodology is that, it can be automated hence, less time is required to carry a particular task and more efficient if the task is more complex and if it is a larger scale task. The proposed methodology showed a very satisfactory result. The purpose of this thesis is to implement an automated database schema matching engine in addition to research and study the techniques and methodologies that is used for schema matching. Keywords: Schema Matching, Individual Match, Multiple Matchers, Convolutional Neural Network