The article presents INarIG, a new model for Word-Level Auto Completion (WLAC) in computer-aided translation. WLAC predicts a target word given a source sentence, translation context, and a human-typed character sequence. Previous models either exploited contextual information or disregarded the dependencies from the right-side context. INarIG, however, structures the human-typed sequence into an Instruction Unit and employs iterative decoding with subwords to fully utilize input information. This approach proves more effective with low-frequency words and significantly improves prediction accuracy.
Publication date: 1 Dec 2023
Project Page: Not Provided
Paper: https://arxiv.org/pdf/2311.18200
