Research on Short Text Similarity Calculation Method for Power Intelligent Question Answering

With the development of artificial intelligence, the question answering system has penetrated into various industries and has become an important production factor. In the electricity field, problems such as the diversification of power equipment failures and the complicated terminology of the power industry are challenging the traditional power question answering system solutions. Therefore, it is of great significance to construct a question answering system based on the knowledge base in the electricity field. However, there are two problems to be solved in the question answering system in this field: (1) How to accurately segment the vocabulary (2) How to effectively match the sentence similarity. To solve the above problems, this paper proposes an algorithm model of cosine similarity combined with TF-IDF. First, add a custom electricity power dictionary in the word segmentation stage, secondly use the space vector model (VSM)-based TD-IDF algorithm for vectorization, and finally, use cosine similarity degree to perform similarity comparison. This method is verified on the electricity power question answering data set, and compared with the LDA model, TF -IDF algorithm and LSI model respectively. The experimental results show that the accuracy of the method proposed in this paper reaches 75.8%, which is significantly better than the other three. It proves that the research model can accurately match user questions, effectively reduce labor costs, and help electric power workers better solve the problems encountered in their work.