An Enhancement of the Jaro-Winkler Fuzzy Searching Algorithm Applied in Library Search Engine

Abstract

The Jaro-Winkler algorithm is widely used for approximate string matching, offering reliable similarity calculations between two strings. However, its performance declines with increasing string length due to bias against longer strings and its reliance on prefix similarity, which neglects significant suffix matches. This paper presents an Enhanced Jaro-Winkler algorithm that addresses these challenges by integrating a Rabin-Karp Rolling Hash – inspired technique and applying suffix weights to balance the prefix bias. Experimental evaluations using 100 words commonly found in book titles demonstrate the enhanced algorithm’s robustness across varying fuzzy match thresholds (0.7, 0.8, and 0.9). Unlike the traditional algorithm, where higher thresholds reduce match accuracy, the enhanced algorithm consistently achieves 100% accuracy in identifying titles regardless of query position or threshold. Additionally, it showcases superior performance by improving the quality and quantity of retrieved results by a significant number of titles compared to the traditional approach. These advancements highlight the algorithm’s potential for improving search performance in applications requiring precise and flexible string matching.