Research on English Chatbot Design for Civil Aviation Service

To further improve the service level of the civil aviation hall, an English chat robot for civil aviation service is proposed, and optimization design is performed on the language generation module and response module of the robot. Where, the Seq2Seq model was combined with the attention mechanism to construct the language generation module. The TF-IDF algorithm was combined with semantic similarity calculation and deep scoring model to realize the generation of chat responses. The results show that compared with the language body generation model before improvement, the proposed language generation model generates longer and more diverse responses. Compared with other response retrieval methods, the retrieval effect of the retrieval method proposed in this paper is better, reaching 0.4766, 0.5012 and 0.2799 in the three evaluation indexes of P@1, Hit@3 and MRR, respectively, which are significantly higher than other retrieval methods. Through the above results and the manual scoring results, it can be concluded that the designed English chatbot can give rich and relevant responses in the conversation scene, which can bring a good experience to passengers in the civil aviation hall and has certain practical application value.