- From: Martynas Jusevičius <martynas@atomgraph.com>
- Date: Fri, 19 Sep 2025 17:14:31 +0200
- To: Joris Lemay <joris.lemay@ehealtho.com>
- Cc: Sebastian Samaruga <ssamarug@gmail.com>, W3C Semantic Web IG <semantic-web@w3.org>, W3C AIKR CG <public-aikr@w3.org>
Hi all, I recently published a project called Web-Algebra that bridges Semantic Web and GenAI. It does not consume non-semantic APIs however. On the contrary, it provides composable operations for RDF Linked Data and SPARQL sources. The operations can be executed in two ways: * as MCP tools (that get executed sequentially by LLMs) * as a new DSL that composes the operation calls into JSON 'bytecode' that is executed by Web-Algebra The operations include SPARQL query execution, read/write Linked Data requests, and LinkedDataHub data management. This way, Web-Algebra allows expressing complex semantic workflows declaratively using a JSON representation. This enables one-shot execution: instead of the LLM making multiple sequential MCP tool calls, it compiles the entire workflow into JSON that Web-Algebra executes atomically in one shot. https://github.com/AtomGraph/Web-Algebra I think you might find it interesting :) Best, Martynas atomgraph.com On Fri, Sep 19, 2025 at 4:33 PM Joris Lemay <joris.lemay@ehealtho.com> wrote: > > Dear Sebastian, all, > > > > Your post on GenAI + Semantic Web for EAI and BI made me think of a related approach from SAP that might interest this group. > > > > SAP has published a short course on API discovery using knowledge graphs that link RDF/OData API metadata to business processes, enabling agents to not only find the right APIs but also execute them in the right order: > > https://www.deeplearning.ai/short-courses/knowledge-graphs-for-ai-agent-api-discovery/ > > > > Key points from the course: > > APIs are ingested into a knowledge graph and extended with process data. > Semantic retrieval reduces the API search space, while process edges define dependencies. > Agents can then select APIs and follow correct process sequences (GET, POST, etc.). > > > > Looks like a solid complement to your exploration of GenAI-leveraged Semantic Web infrastructure for enterprise integration. > > > > Best, > > Joris > > > > <!--This email was generated with AI--> > > > > From: Sebastian Samaruga <ssamarug@gmail.com> > Date: Wednesday, 17 September 2025 at 21:46 > To: W3C Semantic Web IG <semantic-web@w3.org>, W3C AIKR CG <public-aikr@w3.org> > Subject: Semantic Web + GenAI > > Wondering if there is place for some attempts to outline what could be a GenAI leveraged Semantic Web infrastructure for Enterprise Application Integration (EAI) and Federated Business Intelligence (BI). > > The links in this post are some very early drafts of what could be achieved so far: > > https://sebxama.blogspot.com/2025/08/eai-framework.html > > This is a "container" index document where I'll be putting conclusions from the previous blog post links contents: > > https://github.com/sebxama/sebxama/raw/refs/heads/main/Index.docx > > I'm sending it also as an attachment. Looking for corrections / collaboration. Thanks in advance, > > Sebastian.
Received on Tuesday, 23 September 2025 05:12:24 UTC