Automobile Maintenance Case Matching Method Based on the Continuous Space Language Model

Abstract

This paper proposes a method for automobile maintenance case extraction based on a continuous language model to improve the context understanding accuracy and reduce the result redundancy of the traditional keyword matching technology in the car maintenance case diagnosis process. First, the SimHash algorithm is used to find the candidate case diagnosis result sets. Then, a deep learning technology word embedding is used to create a distributive map of the keywords and the candidate case sets and queries. Finally, a continuous space language model-based inference algorithm is developed to produce more accurate matching results between the keywords and the candidate case sets. Experiments carried out on vehicle maintenance case data prove that the continuous language model can effectively improve the case matching accuracy.