User Preference Model-Based Sentiment-Semantic Corresponding Method for Korean Text Translation

In order to solve the problems of poor matching accuracy and long time-consuming in traditional Korean text translation emotional semantic matching methods, this paper proposes research on a Korean text translation emotional semantic matching method considering a user preference model. Build a user preference model, input that obtained emotional space position perception vector into the user preference model, correcting the translation semantic result according to the user’s emotional preference, determining the freshness of keyword features, calculating the freshness weight, generating user preferences, and calculating the similarity of interest vectors among users, on this basis, obtaining a Korean text data set, and using Word2Vec model to map individual words in natural language corpus to vectors to extract Korean text translation features; By inputting word vectors into DPCNN, depth features are extracted to classify text emotions. The semantic matching model of translated text is constructed by sentence representation, and the emotional space position perception vector is introduced to deeply match the emotional semantics of Korean translated text, which makes the semantic matching result more accurate. The experimental results show that the semantic matching accuracy of Korean translation text under this method is 99.6%, and the matching time is 3.6s, which can effectively improve the accuracy of emotional semantic matching and further improve the translation effect of Korean text.