AI时代自然语言处理的驱动力及其逻辑语义学  

Driving Forces of Natural Language Processing and Logical Semantics at the Age of ArtificialIntelligence

作　　者：胡玥 姚从军 HU Yue;YAO Cong-jun(Biquan Academy,Xiangtan University,Xiangtan,Hunan411105,China)

机构地区：[1]湘潭大学碧泉书院,湖南湘潭411105

出　　处：《贵州工程应用技术学院学报》2024年第4期59-66,共8页Journal of Guizhou University Of Engineering Science

基　　金：湖南省社会科学基金一般项目“基于多模态组合范畴语法的汉语研究”,项目编号:23YBA081。

摘　　要：大数据驱动的自然语言处理NLP运用了深度学习方法,使得自然语言处理的性能得到飞速提升。但是它不是在理解的基础上进行处理,因而在处理包含人类知识的复杂语义时显得非常脆弱。攻克人工智能的难关在于使机器真正理解语言,计算机如何表征、分析和理解自然语言的语义是人工智能的关键问题。逻辑语义学基于规则的理性主义取向能够刻画语言构造的无穷机制,但没有面向自然语言的大量真实文本,不能满足人工智能处理自然语言的需要,因而主流NLP目前不愿关注跟自然语言实际情况相对脱节的逻辑语义学研究;经验主义的NLP语料库无法解释语言构造的无穷机制,但能够满足人工智能的实用需要,逻辑语义学应该吸纳经验主义之长,实现理性主义方法与经验主义方法融合和互补。Big data-driven natural language processing NLP uses deep learning methods to make the performance of natural language processing increase rapidly.But it does not process on the basis of understanding,so it is not very fit to deal with complex semantics containing human knowledge.The difficulty in overcoming artificial intelligence lies in making the machine truly understand the language.How computers represent,analyze,and understand the semantics of natural language is a key issue for artificial intelligence.The rule-based rationalist orientation of logical semantics can characterize the infinite mechanism of language construction,but without a large amount of real text for natural language,it cannot meet the processing of artificial intelligence.Thus the mainstream NLP is currently reluctant to focus on logical semantics research that is relatively disconnected from the actual situation of natural language.The empirical NLP corpus cannot explain the infinite mechanism of language construction,but it can meet the practical needs of artificial intelligence.Logical semantics should absorb the strengths of empiricism and realize the integration and complementation of rationalist methods and empiricist ones.

关 键 词：自然语言处理 大数据 大知识 逻辑理性主义 经验主义 

分 类 号：B81[哲学宗教—逻辑学]