Towards a universal language of concepts: A survey
人类能从稀疏数据中学习并泛化新概念,因为知识以丰富的结构格式表达。本文提出程序是概念通用表示的强有力候选者。文章综述了使用程序作为概念表示的计算模型,并评估其对构建通用表示语言所做的贡献。
EVENT DOSSIER
On September 7, 2026, arXiv published a review article titled “Towards a universal language of concepts: A survey”. This study systematically reviewed the current status and challenges in constructing a universal concept language in the field of artificial intelligence, aiming to explore theoretical frameworks and technical approaches for achieving cross-modal and cross-domain unified semantic representation, thereby providing basic theoretical support for the development of multimodal large models and general artificial intelligence.
人类能从稀疏数据中学习并泛化新概念,因为知识以丰富的结构格式表达。本文提出程序是概念通用表示的强有力候选者。文章综述了使用程序作为概念表示的计算模型,并评估其对构建通用表示语言所做的贡献。