Assisted translation method

Abstract

The invention provides an assisted translation method integrating translator input, a memory bank fragment / term bank and real-time preview. The assisted translation method comprises the following steps: 1, analyzing and splitting an editable document uploaded by a translator into independent sentence segments by taking sentences as units; 2, associating the independent sentence segments with a memory library and a term library, and extracting memory library segments in batches so as to be fused into recommended translations for translators in an interactive translation link; 3, displaying documents in the analyzed independent sentence segments in an HTML format which can be previewed and edited in a browser, and meanwhile, subjecting sentence pairs in the analyzed independent sentence segments to word segmentation for term library matching; 4, matching the parsed sentence segments and word segmentation results in an associated memory bank and / or term bank, and marking sentences meeting preset conditions; and 5, returning the translated text by a machine translation engine.

Description

technical field [0001] The invention relates to an auxiliary translation method which integrates translator input, memory store fragments / term bank and real-time preview. Background technique [0002] Neural network machine translation makes the automatically translated translation more meaningful for the translator. The more popular translation method in the market is the post-translation editing mode (the first type of solution)—that is, allowing the translator to directly modify the automatically translated translation. Another type of solution is interactive translation based on the translator's input segment re-decoded to generate a translation (the second type of solution). The machine recommends translations in real time based on the translator's input, and does not provide memory segment / term base fusion and editor preview. [0003] The disadvantage of the first type of solution is that the machine-translated text directly occupies the translation input box, which li...

AI Technical Summary

Problems solved by technology

[0003] The disadvantage of the first type of solution is that the machine-translated text directly occupies the translation input box, which limits the translator's translation ideas and easily misleads the translator, causing the translator to only make minor repairs to the machine translation results, and the content modified by the translator cannot realize the machine model. Quantitative learning

The disadvantage of the second type of solution is that it does not integrate the translator's memory fragments and term bases, as well as the full-text preview information. When interactive translation prompts, only the recommended translation based on machine translation is provided, and the memory fragments corresponding to the current sentence cannot be integrated. Terminology recommends translations in real time, resulting in low recommendation accuracy and limited efficiency improvement for translators

Images