Scholarlink

<p>The wide use of social networks is generating a huge quantity of data, structured and unstructured. With increased data volume and complexity, traditional databases became less efficient, which led to the development of NoSQL, a new generation of Database Management Systems. Despite the typical schema-optional nature of NoSQL databases, researchers have conducted studies on their design and emphasized the need for modeling techniques tailored to NoSQL data management. Graph-oriented databases, as a part of the NoSQL ecosystem, are gaining popularity because of their querying efficiency for complex connected data. Multiple aspects must be considered when modeling a graph-oriented database, generally a NoSQL database, including the relevant domain data and how these data will be accessed. Although several academic works have proposed modeling processes for graph-oriented databases, most of them do not rely on metamodeling to automate the design process and often leave important steps insufficiently specified. This study addresses the limitations observed in existing graph database design approaches by proposing a modeling guide that considers key requirements of graph-oriented database design. The proposed approach relies on Model-Driven Architecture to structure the design process and automate the generation of multiple models. By combining metamodeling, query-driven design, and formal specification, the approach improves model quality and maintainability while explicitly integrating performance considerations into the schema design process. We conducted an experiment on graph databases containing millions of nodes and relationships. Experimental results indicate that the proposed design achieves query execution time reductions ranging from 17% to 99% compared to the baseline schema, and remains more efficient than other representative state-of-the-art design approaches, while introducing an additional space overhead of approximately 62%. This balance highlights the method’s practicality and its suitability for large graph databases.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

GRAPHDESIGN: a model-driven approach for a query-oriented graph database design

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月3日

查看全文
  • Ahmed Dourhri,
  • Mohamed Hanine,
  • Hassan Ouahmane,
  • Ernesto Bautista Thompson,
  • Daniel Gavilanes,
  • Imran Ashraf

摘要

The wide use of social networks is generating a huge quantity of data, structured and unstructured. With increased data volume and complexity, traditional databases became less efficient, which led to the development of NoSQL, a new generation of Database Management Systems. Despite the typical schema-optional nature of NoSQL databases, researchers have conducted studies on their design and emphasized the need for modeling techniques tailored to NoSQL data management. Graph-oriented databases, as a part of the NoSQL ecosystem, are gaining popularity because of their querying efficiency for complex connected data. Multiple aspects must be considered when modeling a graph-oriented database, generally a NoSQL database, including the relevant domain data and how these data will be accessed. Although several academic works have proposed modeling processes for graph-oriented databases, most of them do not rely on metamodeling to automate the design process and often leave important steps insufficiently specified. This study addresses the limitations observed in existing graph database design approaches by proposing a modeling guide that considers key requirements of graph-oriented database design. The proposed approach relies on Model-Driven Architecture to structure the design process and automate the generation of multiple models. By combining metamodeling, query-driven design, and formal specification, the approach improves model quality and maintainability while explicitly integrating performance considerations into the schema design process. We conducted an experiment on graph databases containing millions of nodes and relationships. Experimental results indicate that the proposed design achieves query execution time reductions ranging from 17% to 99% compared to the baseline schema, and remains more efficient than other representative state-of-the-art design approaches, while introducing an additional space overhead of approximately 62%. This balance highlights the method’s practicality and its suitability for large graph databases.