A Comparative Evaluation of Knuth-Morris-Pratt Algorithm and Large Language Models with Retrieval Augmented Generation

Resume screening is pivotal in modern recruitment, determining which candidates advance in the hiring process. As companies receive many applications, especially for high-demand roles, efficient and accurate screening methods are essential to avoid overlooking potential talent. This paper examines two advanced methods: the Knuth-Morris-Pratt (KMP) algorithm, known for its fast and efficient pattern matching, and Large Language Models (LLMs) with Retrieval-Augmented Generation (RAG), which provide context-aware resume analysis. Recent studies have focused on leveraging artificial intelligence and machine learning to enhance recruitment processes. LLMs show significant promise in handling unstructured text data, understanding nuanced qualifications, and improving hiring outcomes. This paper aims to build on these advancements by directly comparing the performance of KMP and LLMs with RAG in key areas such as processing time and accuracy. While KMP excels in speed, it cannot interpret deeper meanings, whereas LLMs with RAG handle unstructured data more effectively but demand higher computational resources. Ultimately, this study finds that LLMs with RAG outperform KMP in complex resume screening tasks, offering better accuracy and scalability, albeit with greater computational costs.