inno.trek

「Where innovation never stops.」

Data&AI技术趋势 2025

背景 人类社会历史是螺旋前进的,科技史亦然。从1950s提出Artificial Intelligence概念至今,AI技术已经经历了多次春、冬季交替演进(AI Winter)。上一代AI技术(1970s-1980s)是基于符号(symbolic)的计算系统,以LispMacine、专家系统、日本第5代计算机系统等为代表。这类技术在知识表示、逻辑推理上有优势,但是不具备模式识别能力。此后基...

Why self-reference matters in AI

Background Self-reference, one of the most important concepts for understanding human consciousness, is both a fundamental aspect and a powerful tool for advancing future AI systems. Its profound ...

Why self-explanation matters in AI

Background The success of LLMs has triggered a new wave of prosperity of AI researches. The market, supply chains and end-users have been educated in a unprecedented depth about AI capabilities in...

Ontology Framework

概述 本文总结当前主流的语义网络(知识图谱)领域的本体描述框架,包括理论,语言和系统。 其中本体论介绍相关理论发展,以及大家对其形而上学的争论;本体工程介绍若干基础和 领域的本体概念集;本体语言介绍典型的知识表示语言及其发展过程。最后展望在大语言 模型变革下的本体论技术发展趋势。 简史 符号知识系统的发展,从最初的通用问题解决器GPS 到最近的语义网络和知识图谱,经历 了多次的基础理论...

Zen of Computer Language: Lisp

S-expr and evaluation A lisp program is a set of s-expressions. The basic of s-expressions are lists and atoms. An atom is a non-list or the empty list. Lists are delimited by parentheses and cont...

Zen of Computer Language: JS

背景 我比较喜欢研究计算机语言理论。所有的计算机语言来自于历史上诸多计算机学者和工程人 员的努力。每种语言都有生而为之的特征。每年都有诸多计算机语言诞生,而追寻每种语言 背后的故事以及了解它们的与众不同的特点,很有意思。这是What X surprises me系列文 章之一,记录某种语言的设计特点在我的认知中带来某种惊讶或惊喜。有点钟书先生《管锥 编》异曲同工之意。这篇有关Javasci...

Knowledge Representation (Part I)

Intro An efficient representation of knowledge/logics lies at the core of knowledge computing engines. This requirement is also one of vital pieces of future artificial general intelligence (AGI)....

Four Ways of Knowledge Representation

Knowledge representation involves representing the key concepts and relations between the decision variables in some formal manner, typically within a framework suggested by an expert systems shell...