Development of an Enhanced BP Neural Network-Based English Translation Evaluation System Integrating GLR Algorithm for Automated Judgment

The study proposes an enhanced English Translation Scoring System (ETSS) using an improved Generalized Maximum Probability Ratio (GLR) algorithm and an expanded BP neural network to improve translation evaluation accuracy. The system incorporates a three-phase method for text feature extraction, optimization, and fusion. Key improvements include wavelet packet decomposition for feature analysis and context-aware modifications for resolving ambiguities. The ETSS demonstrates over 95% accuracy in identity detection and an overall reliability of 92.3%, confirming its effectiveness for automated English translation assessment.