LLMs are starving for knowledge graphs. Raphael Troncy was pointing out that many LLM company crawlers are constantly visiting their KGs. Some crawlers even perform explicit SPARQL queries on the KGs.
@lysander07 The image shows a presentation slide titled "LLMs are starving for KGs" (Large Language Models are starving for Knowledge Graphs). The slide is projected onto a screen and features a list of crawlers visiting various Knowledge Graphs (KGs), including OpenAI, ByteDance, Apple, Meta AI, Anthropic, Microsoft, DuckDuckGo, CommonCrawl, Amazon, and Perplexity. Each crawler is associated with a specific KG, and the number of requests made to each KG is listed. For example, OpenAI has made 3,430,585 requests to data.doremus.org, while Perplexity has made 48 requests to yang.eurecom.fr. The slide also notes that the crawlers are mostly visiting endpoints like /describe/, /fct/, and /sparql. The presenter, a man in a light blue shirt, stands at a podium with a microphone, addressing an audience. The background of the slide is light blue, and the text is in black, with a blue sidebar on the right side containing additional data. The bottom right corner of the slide displays the logo of EURECOM, a research institute.
Provided by @altbot, generated privately and locally using Ovis2-8B
Energy used: 0.247 Wh
@lysander07 i've actually started turning many of my git repos into Obsidian vaults so that the MCP service in Pieces and other desktop applications can draw from those sources via MCP or RAG and it's like a parade because candy is flying everywhere
@lysander07 oh yeah I see the same on the @SIB #SPARQL endpoints. Massive crawling by LLM crawlers.