Automatic Extraction of Operator Industry Specification Terms Based on Multi-Combination Unsupervised Methods

The operator domain terminology extraction algorithm can extract the core vocabulary terms of the operator industry to understand the content of the operator's vertical scenarios quickly. Traditional supervised machine learning, sequence annotation methods require a large number of manual annotations, especially for domain-specific data, the annotators also need to have specialized knowledge and the cost is high. To address the above problems, this paper proposes to use a blend of unsupervised extraction methods for the automatic construction of the operator industry thesaurus by combining the statistical method TF-IDF and large model-based prompt extraction and other methods to obtain the operator industry terminology unigram, bigram and trigram, respectively, a total of 16,539 words, the accuracy of unigram automatic extraction reaches more than 95%, which can be used for operator industry knowledge search, Q&A and other fields.