Power intelligent question answering customer service robot short text specific semantic mining

In short texts, there is often insufficient context information, so it is necessary for robots to have reasonable reasoning and understanding of context. For the understanding of professional terms, abbreviations and specific context scenarios in the power industry, it is necessary to dig deep into the semantics to make a correct answer. Therefore, this paper proposes a short text specific semantic mining method applied to the power intelligent question answering customer service robot. Capture the binary classification corpus of power customer service voice mining, and design the specific flow of this process. Fuzzy C-Means clustering (FCM) algorithm and immune single-parent genetic algorithm are combined to construct a short text specific semantic mining method for customer service robots. The experimental results show that for different sub-data sets, the average F1 value of the design method is higher than 0.9, indicating that the design method has a good performance of comprehensive short text specific semantic mining. The standard measure of semantic segmentation of the design method is in the range of 0.85-0.90 when the short text specific semantic delivery is high and low. The results show that the method has wider applicability and better performance.