Intelligent translation method for legal provisions based on multilingual semantic model

Abstract

The present invention discloses an intelligent translation method for legal provisions based on a multilingual semantic model, including: obtaining initial semantic difference features from legal provisions through a multilingual semantic network, performing preliminary comparison for different legal systems and language systems to obtain a quantitative representation of conceptual meaning deviations; obtaining cross-language equivalents of relevant legal concepts from a knowledge integration database based on the obtained refined semantic structure, performing semantic alignment processing for translation accuracy requirements to determine a final semantic equivalent translation draft; if there are residual deviations in the determined semantic equivalent translation draft, adopting a feedback loop mechanism to trace back to expert knowledge resources, obtaining supplementary multi-party collaborative annotation data to acquire a revised translation output version, conducting complete semantic verification on the entire legal provisions, and outputting the final accurate translation result after judging that the verification is passed by comparing the depth of semantic understanding between the source and target versions. The present invention significantly improves the accuracy and efficiency of legal translation.

Images