Welcome Post

Welcome to the AI KR CG

This Post gives some background and points to work in hand. Written for new members but works as a refresher for all participants

Members are welcome to post to the AI KR CG Mailing list, introduce themselves and share their work. Links to talks, papers, events and meetings are posted on the list.

The WIKI is where we try to share meaningful resources These are generally diagrams, mindmaps, papers and vocabularies. May need curation.

Background and Evolution of the AI KR Community Group

The initial mission of the AI KR Community Group (CG) was defined in 2018, when the group was first established. While the core vision remains relevant, the scope has evolved significantly over time. Although we have not yet been able to update the CG landing page to fully reflect this evolution, several related wiki pages have been revised to better represent the expanded direction of the group.

The Original Motivation

Between roughly 2014 and 2018, machine learning–based AI became a primary driver of technological innovation. During this period, increasing attention was given to the “black box” nature of many machine learning systems — particularly the lack of explainability, interpretability, and replicability of models.

The AI KR CG was founded on the understanding that **Knowledge Representation (KR)** could play a critical role in addressing these challenges. The premise was that structured knowledge models, ontologies, and semantic technologies could help:

Improve understanding of AI systems
Enhance explainability and transparency
Support reproducibility and interoperability

The Shift Toward Large Language Models

With the emergence of large language models (LLMs), the relevance of KR — especially **natural language–based knowledge representation** — has become even more evident.

LLMs:

Accept natural language prompts as input
Produce natural language outputs
Serve as interfaces to broader AI and machine learning systems

Today, language models are increasingly used as intermediary interfaces to AI tools, applications, and environments that may not themselves have native LLM capabilities. In this new paradigm, KR supports:

Structuring prompts
Designing controlled vocabularies
Defining semantic constraints
Creating transparent interaction layers
Explaining AI-driven architectures

In short, KR helps us both **understand how AI architectures are evolving** and **interface with them effectively through structured natural language interaction**.

Current Scope

The primary focus of the AI KR CG is therefore:

> To understand how natural language interfaces are being integrated into AI systems at multiple architectural levels, and how knowledge representation can support this integration in a principled, interoperable, and trustworthy way.

Within this scope, there is ample room for dialogue across a wide range of related topics that may interest members.

Current Use of the Mailing List

To date, the AI KR CG mailing list has primarily been used to:

Raise general discussion topics
Share relevant resources
Exchange references and ideas

While valuable, this usage has not yet fully activated the group’s potential for structured collaboration and deliverables.

Strategic Direction: Supporting Broader W3C Work

Looking ahead, the group aims to actively support other W3C communities and working groups in developing best practices and standards related to:

Agentic Ontologies
Interoperability
Digital Forensics
Smart Voice Agents
Privacy and Trust in AI
Web AI
Agent Automation
WebMCP vs. MCP
Agent-to-Agent (A2A) communication

Potential outputs may include:

Use case collections
Controlled vocabularies
Ontology patterns
Model card generator specifications
Semantic interface specifications
Architectural design patterns

These efforts would aim to ensure that AI systems integrating natural language interfaces are interoperable, explainable, and aligned with web standards.

Call for Engagement

To make this work relevant and impactful, increased participation from CG members is essential.

We should consider:

Forming focused task groups around specific themes
Hosting structured thematic discussions (e.g., monthly deep dives)
Developing shared working drafts
Identifying concrete standardization gaps
Collaborating with adjacent W3C groups
Running small pilot projects or demonstrators

Open Questions to Members

How can we best encourage active involvement and sustained collaboration within the AI KR CG? What are participants priorities, given the premises outlined above? Please get in touch with the Chair if you d like to take discuss some ideas or take responsibility for a deliverable *Use cases, specifications, recommendations

Some possible approaches:

Clear short-term deliverables
Contributions and collaborations with funded projects and research
Joint position papers
Inter-group workshops
Meetings to discuss possible Draft specifications

-