Shared Word Embedding Space Modeling Method Based on Orthogonal Projection

With the continuous development of computer technology, machine learning has been applied in more and more fields. However, the application of word embedding technology in bilingual Chinese and English still needs to be developed. In this paper, we propose a model construction process based on orthogonal projection, and analyze the validity of the model from multiple perspectives. We carry out word sense similarity experiments and word analogy experiments for the quality of single language in the model, and cross-language text similarity experiments for different linguistic quality in the model. Through the analysis of the experimental results, it can be proved that the proposed shared word embedding space model achieves good results compared with the traditional word embedding model, and the effect of the model achieves the desired purpose.