Intelligent accounting question-answering robot based on a large language model and knowledge graph

Abstract

In the wave of digital transformation, enterprises have an increasing demand for fast and accurate access to financial information. The conventional accounting service model often relies on manual operations, which are not only inefficient but also susceptible to errors. This study aims to design an intelligent accounting question-answering robot based on a large language model and knowledge graph. To build a complete knowledge graph, this study uses the attention mechanism and convolutional neural network to build a connection prediction model and completes the accounting question-answering knowledge graph. After that, the bidirectional gated loop unit is used to improve the large language model so as to further improve the correlation between knowledge and explore potential information. The results denoted that the developed method had a question-answering accuracy of 94.6%, and the answers covered 95.2% of the domain range. The response time was only 120 ms, which was faster than other models and enhanced the user experience. Moreover, the user satisfaction score of Model 1 was 9.2 points. It is expected that the designed bot will be helpful for enterprises to obtain quick financial information and improve accounting service efficiency.

Keywords

• large language model
• knowledge graph
• intelligent accounting
• question answering
• robot