Comparative Performance Analysis of String Matching Algorithms and Data Matching Frameworks Using Python Libraries on Academic Datasets

This study conducts a comparative analysis of diverse string matching algorithms (Jaro-Winkler, Edit Distance, Jaccard, Hamming) and data matching frameworks (Splink, RLTK, etc.) implemented using Python libraries on degraded academic datasets. Methodologically, performance was quantified via precision, recall, and F1-score across varying degrees of data corruption. Findings indicate that no single method achieves universal optimality. Jaro-Winkler and the Splink framework demonstrate superior precision, minimizing false positives. Conversely, Edit Distance and the RLTK framework yield maximum recall, capturing a broader range of potential matches but at a higher risk of false positives. The implication is that reliable entity resolution necessitates a tiered, adaptive pipeline. This integrated approach balances the trade-offs between precision and recall, supporting consistent, cost-effective, and reliable entity resolution across heterogeneous academic records.