Efficient Top-k Keyword Search in Relational Databases Considering Integrated Candidate Network

Abstract

Efficiently navigating vast datasets without requiring query language expertise is crucial in the era of Big Data. Keyword search in relational databases offers a promising solution, but many existing methods struggle with datasets of moderate size, such as one million tuples. This paper introduces a novel approach using an Integrated Candidate Network (ICN) to enhance query responses by reducing redundant operations and pruning non-promising candidate networks. Unlike approaches focusing on Large Language Models (LLMs) for unstructured data, our method uniquely optimizes structured data environments. Experimental evaluations across diverse real-world databases demonstrate significant enhancements in query efficiency and effectiveness. This research contributes to advancing keyword search in relational databases by leveraging structured data principles to address current limitations.

Keywords

• Effectiveness
• efficiency
• integrated candidate network
• keyword search
• pruning
• relational database