Research on Financial Service Robot Chat System Based on Text Matching

In order to provide users with more accurate financial knowledge in financial services, a financial services chat question answering robot system based on text matching model is proposed. On the basis of determining the overall architecture of the chat question answering robot system, ALBERT model is taken the basic text matching model. At the same time, the corresponding improvement is conducted on ALBERT model and it is combined with Attention-CNN structure, so as to further improve the quality of text matching. The experimental results reveal that compared with other types of text matching models, the text matching model constructed in this paper performs better in accuracy and F1 score, reaching 80.1% and 82.5%, respectively, with high accuracy. Building the constructed text matching model in the chat question answering robot system can effectively improve the human-computer interaction effect of the financial knowledge question answering process, which verifies the effectiveness of the matching model. Therefore, the constructed financial service chat question answering robot system based on text matching model has good performance, can effectively improve the interaction effect of the question answering process, provide users with accurate financial related knowledge, and has certain practical application value.