Generating Effectiveness Entities of Patent Technology Based on ChatGPT+Prompt
[Objective]This paper constructs a new model that automatically identifies and extracts patent technology and function entities.It constructs a matrix for high-quality technology and function.[Methods]We utilized ChatGPT+Prompt to extract patent technical efficacy entities and recognize,extract,and generate technology and function words with technology-function binary groups.[Results]The proposed method recognized and generated patent technology and function entities in four domains and three languages.Our method can generate technology-function binary groups more accurately in the cross-domain,cross-language,and prompted sample size comparisons.The model yielded the highest ROUGE values with the electronic automobiles and English patents.Giving One-Shot will significantly improve the model's cross-domain and cross-linguistic performance.[Limitations]The proposed method lacks standards for prompts,generates duplicated contents,and needs multi-round Q&A.[Conclusion]The proposed method effectively reduces the labor cost and task threshold of technology and function entity generation.It expands the application scenarios of AIGC and releases the potential of ChatGPT in patent document exploration.
Patent Technology Function MatrixTechnology WordsFunction WordsEntity RecognitionGenerative ModelsChatGPTPrompt
NETL
NSTL
万方数据
