中圖分類號(hào)：TP391.1 文獻(xiàn)標(biāo)志碼：A DOI: 10.16157/j.issn.0258-7998.256942
中文引用格式： 楊嘉佳. YJJFA：一種數(shù)據(jù)驅(qū)動(dòng)的高性能正則表達(dá)式匹配算法[J]. 電子技術(shù)應(yīng)用，2026，52(4)：102-107.
英文引用格式： Yang Jiajia. YJJFA: a data-driven high-performance regular expression matching algorithm[J]. Application of Electronic Technique，2026，52(4)：102-107.

YJJFA： a data-driven high-performance regular expression matching algorithm

Abstract： Under the background of the artificial intelligence era, regular expression matching technology plays a crucial role, particularly in data cleansing and data extraction domains, where it provides technical support for high-quality data processing required by large language model training. However, traditional regular expression matching algorithms suffer from performance bottlenecks that limit their application scope. To address this challenge, this paper proposes a data-driven high-performance regular expression matching algorithm, denoted as YJJFA algorithm. The algorithm reduces the number of input characters that need to be processed in the state transition table by partitioning it into optimal trusted regions and untrusted regions, while leveraging non-memory-access vector comparisons of the untrusted character set (UBCS) to achieve low-time-cost processing of trusted characters. Experimental results show that the YJJFA algorithm achieves a throughput rate of 17.88~53.81 Gb/s on L7filter rules, representing an order-of-magnitude performance improvement over conventional DFA implementations.

Key words : regular expression matching；trusted zones；high-performance；data cleansing