基元潜部特征元挖掘的智能方法研究  

Research on intelligent methods for latent features mining of basic element

作　　者：张丽芳 李兴森 ZHANG Lifang;LI Xingsen(Institute of Extenics and Innovation Methods,Guangdong University of Technology,Guangzhou 510006,China)

机构地区：[1]广东工业大学可拓学与创新方法研究所,广东广州510006

出　　处：《智能系统学报》2025年第2期457-464,共8页CAAI Transactions on Intelligent Systems

基　　金：国家自然科学基金项目(72071049);广东省自然科学基金项目(2024A1515011324).

摘　　要：潜部特征元构建是可拓学基元理论的重要研究内容,潜在信息挖掘对解决问题和激发创新思维至关重要。为了解决目前人工识别基元潜部特征元效率低、覆盖面窄和数量不足的问题,研究基元潜部特征元显化理论与人工智能算法实现的结合点,提出挖掘基元潜部特征元的流程化、系统性方法,使用爬虫技术收集目标基元对象的相关信息,清洗噪音数据并从句子中挖掘构成特征元的名称和描述,用概率统计的定量方法筛选潜部特征元并通过Python代码实现智能挖掘功能,最后通过案例对比分析进行论证。研究结果能有效提高基元潜部特征元的识别效率和智能化水平,对从复杂多变的语料句法中进行语义概括也有一定的参考作用,为进一步提高特征名称及其量值智能提取的精确性积累训练集,促进可拓展型人工智能理论的发展。Latent feature element construction is a key aspect of the basic-element theory of extenics,and mining latent information is crucial for problem solving and fostering innovative thinking.This study explores the integration of the basic-element latent feature element manifestation theory with artificial intelligence algorithms to address the current problems of low efficiency,narrow coverage and the insufficient number of manually identified basic-element latent feature elements.A process-oriented,systematic method for mining latent feature elements of basic elements is proposed.The method involves using crawler technology to collect relevant information regarding target basic-element objects,cleaning noisy data,and mining names and descriptions of constituent feature elements from sentences.A probability statistical approach is then used to filter latent feature elements,with the intelligent mining process implemented through Python code.Finally,a case study comparison is performed to demonstrate the effectiveness of this approach.Research results can notably improve the recognition efficiency and intelligence level of basic-element latent feature elements while also providing valuable insights for semantic generalization from complex and changeable dynamic corpus syntax.Additionally,it contributes to building a training set for enhancing the accuracy of intelligent extraction of feature names and their quantitative values,thus promoting the development of extensible artificial intelligence theory.

关 键 词：可拓学 潜部特征元 特征元 基元理论 人工智能 自然语言处理 大语言模型 可拓智能 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]