Specialized Terminology Consistency Maintenance Algorithm for English-Chinese Translation Combining Hidden Markov Model and Deep Learning

Terminology consistency is a key factor in ensuring the accuracy and professionalism of professional translation, especially in terminology-intensive fields such as medicine and engineering. However, the current mainstream terminology translation methods still have major deficiencies in dealing with terminology ambiguity, contextual semantic drift, and terminology variants, making it difficult to meet the requirements of high-quality professional translation. To solve this problem, this paper proposes a terminology consistency modeling method that combines hidden Markov model (HMM) and convolutional neural network (CNN). This method uses HMM to simulate the state evolution of terms in the translation process, and extracts the contextual semantic features of terms through CNN, thereby achieving accurate modeling and evaluation of the terminology consistency state. We conducted experiments on translation datasets in the two professional fields of medicine and engineering. The results show that this method is superior to traditional alignment methods and single neural network models in key indicators such as terminology consistency accuracy and consistency index. In particular, it shows stronger stability and generalization ability in terms of terminology retention rate and terminology evolution recognition ability. This study not only verifies the significant effect of the fusion strategy of structural modeling and semantic feature extraction in improving terminology consistency, but also provides a general path with theoretical depth and practical application value for terminology consistency modeling, which has broad promotion prospects.