Construction of a Professional Vocabulary Database for Power Transformers Based on Automatic Word Segmentation

Power transformers are essential equipment in power systems for energy conversion and transmission. The construction of a knowledge graph for power transformers can effectively organize relevant knowledge about them. To achieve this, the first step is to build a specialized vocabulary database for the power industry. This paper employs an automatic word segmentation algorithm capable of processing both specialized power-related corpora and non-specialized general corpora to generate an initial vocabulary list. Given the potential for non-relevant words and non-word elements introduced during unsupervised word segmentation, optimization strategies are further introduced, utilizing the product of left and right entropies as a screening criterion to aggregate and purify the vocabulary. By comparing the word segmentation results of specialized and non-specialized corpora, a professional vocabulary database in the field of power transformers has been successfully constructed. This accomplishment not only automates the recognition and accumulation of power industry-specific vocabulary but also lays a solid foundation for the subsequent construction of a knowledge graph for power transformers.