We present FastLexRank, an efficient, scalable, and reproducible implementation of the LexRank algorithm for text ranking. Designed to address the computational and memory complexities of the original LexRank method, FastLexRank significantly reduces time and memory requirements from O(n 2 ) to O(n) without compromising the quality or accuracy of the results. By employing an optimized approach to calculating the stationary distribution of sentence graphs, FastLexRank produces the same results as the original LexRank, but with much greater computational efficiency. Our method also enhances interpretability by ranking posts based on their similarity to the mean embedding-allowing researchers to understand why specific content is considered central. This paper details the algorithmic improvements that enable the processing of large datasets, such as social media corpora, in real-time. Empirical results demonstrate its effectiveness, and we propose its use in identifying central tweets, which can be further analyzed using advanced NLP techniques. FastLexRank supports open, reusable computational workflows via an open-source implementation and standardized embedding models, making it well-suited for reproducible social science research, addressing the growing need for efficient and transparent processing of digital content.

PDF