基于深度学习的跨语言检索技术研究OA

In recent years,the explosive growth of online information has triggered extensive atten-tion to cross-lingual information retrieval methods.Deep learning-based cross-lingual retrieval models possess advantages such as the ability to learn the deep feature patterns of languages and good generali-zation capabilities,but they also face challenges such as poor initial ranking performance and insufficient relevance of top results.To further enhance the accuracy and response efficiency of deep learning-based cross-lingual retrieval,this paper takes the typical BLADE(bilingual lexical and document expansion)model as the foundation,and improves the deep learning-based cross-language retrieval model based on re-ranking principles while enhancing the initial ranking of the BLADE model.Finally,the future devel-opment trends of cross-lingual retrieval are prospected.