Kurt Cagle

📢 Warning: Beware of Predatory Journals Predatory journals exploit researchers and scholars by charging hefty fees for publication without offering proper peer review, editorial oversight, or academic standards. It's essential for academics to recognize these unethical practices to avoid falling victim. 🚨 #phd #journal #publication

10

🏛️ The #CIDOC Conceptual Reference Model (CIDOC-CRM) is an ISO Standard ontology for the cultural domain that is used for enabling semantic interoperability between museums, libraries, archives and other cultural institutions. 📄 In this paper, authors analyze the main processes and tasks, and survey recent works that apply #machine_learning techniques for reducing the costs related to CIDOC-CRM-based compliance and interoperability. https://lnkd.in/gqV7BV8X 💡 Moreover, authors introduce their vision for the given topic, and provide a list of open CIDOC-CRM datasets that can be potentially used for machine learning tasks.

4

This is significant for a number of reasons.Why RAG is so powerful is that it provides the interaction component that an LLM by itself doesn't do. Consider this - while there are some design principles that discourage this usage, if you have a RAG call to an external resource with a prompt that acts as an envelope for passing POST body content to a service, the RAG call is in essence acting to facilitate an agentic mutation event - in effect initiating an action at an end-point. Right now, there are some limitations because RAGs are synchronous, but what happens if you can pass a token back to the RAG engine that can be queried to respond when the state of those tokens change. Aynchronous prompt environments look to me to be the next stage in the process of agentic development. Put another way, right now, when I use ChatGPT, I have to wait for 20-30 second for an image to get generated or an analysis to be performed, at which point I'm blocked. On the other hand, suppose that I pass the pointer of an initiated action to an asynchronous handler that then inserts the action or analysis when the action being handled completes. You can sort of that do that now if you're willing to write your own client, but a true agentic system needs to have this capability natively.

24

2 Comments

Good News: The Topic added the journal /Sustainability/, at present, we received more than 10 papers. 😍 #Spatial Decision Support Systems (#SDSSs) are designed around #geospatial data, models, and analytical tools that collectively support human #planning and decision-making procedures in multiple application areas. These areas are constantly evolving to better address existing real-world challenges and find innovative ways forward such as in enabling and facilitating #urban sustainability. https://lnkd.in/gCY7C3BU

2

https://lnkd.in/ex8H-_h4 is a document on knowledge management that Semantic Arts, Inc. contributed to. Please discuss. As we extend from incorporating the FAIR Principles and knowledge graphs based on data in RDF with explicit semantics described in OWL we need to think about what it all means, and how we can express and distribute this knowledge within and beyond our enterprises

28

13 Comments

Excited about AI's capabilities in data analysis, literature reviews, and manuscript proofreading? Concerned it might replace researchers? Join me and Abhigyan Arun Damita Snow, CAE, ENV SP Penelope Lewis and Stephanie Lovegrove Hansen as we explore 'The Other AI: The Role of Actual Intelligence in the Future of Scholarly Publishing.' Let's separate the hype from reality and discuss how AI can complement rather than replace human expertise. Very encouraging to see great participation already! #SSP2024

40

This is by no means a new or hot take, but an important one: qualitatively data is not just bad quantitative data. In this GovLoop article, I cite Indi Young’s data spectrum, a wonderful meme I saw on UXinsight, and provide rebuttals to some of the common misconceptions on qualitative data. #qualitativedata #mixedmethods #research #uxr #civictech #govcon

58

1 Comment

Librarian of Congress Expands DMCA Exemption for Text and Data Mining DENYING "Class 4: New exemption for computer programs, for purposes of good-faith artificial intelligence (“AI”) trustworthiness research." "The Office received a petition for a new exemption to examine one aspect of these implications—AI trustworthiness. The proposed exemption would cover the circumvention of safeguards and contract terms on online platforms hosting AI models for the purpose of researching biased and other undesirable outputs. After reviewing the rulemaking record, the Register concludes that an exemption is not necessary to allow this socially important research and recommends denying the petition."   https://lnkd.in/eip3cg-v

12

1 Comment

one of the best articles on #graphrag I have seen https://lnkd.in/ehByn-g6

30

🚀 New Blog Alert! 🚀  "The Ghost Stays in the Picture, Part 1: Archives, Datasets, and Infrastructures" by Eryk Salvaggio, a 2024 #Flickr Foundation Research Fellow, is a unique, thoughtful essay. Eryk explores how archives and datasets shape AI-generated images, introducing the concept of "ghosts" that haunt these creations. His view of how the unseen forces embedded in training data influence the final output and reflect on the implications for AI systems is an excellent read. Get more of my thoughts here: https://lnkd.in/dACxuRG4 #AIResearch #DigitalArchives #AIinArt

4

An article on Substack entitled The Death of Data Modeling has aroused heated discussions among data practitioners. According to the article, many of those data practitioners feel that the process is no longer suitable for today’s needs. The author... https://lnkd.in/eJq5VmJg #ArtificialIntelligence #businessintelligence #Data

3

▶ Researchers are developing powerful general-purpose agents by using Foundation Models as modules within agentic systems, but the history of machine learning suggests that hand-designed solutions will eventually be replaced by learned solutions. ▶ Automated Design of Agentic Systems (ADAS) is a new research area that aims to automatically (or, maybe, "automagically?") create powerful agentic system designs, including inventing novel building blocks and/or combining them in new ways. ❓ Why Should You Care❓ ▶ This research demonstrates a promising approach within ADAS where agents can be defined in code and new agents can be automatically "discovered" by a meta agent programming ever better ones in code. ▶ The authors present a simple yet effective algorithm named "Meta Agent Search" that can progressively invent agents with novel designs that greatly outperform state-of-the-art hand-designed agents, while maintaining superior performance even when transferred across domains and models. AI building AI building AI...afraid yet? #agenticai #ai #llm Paper: https://lnkd.in/g77cnY7f

7

3 Comments

Brains and #neuromorphic #systems learn with local learning rules in online-continual learning scenarios. Designing neural networks that learn effectively under these conditions is challenging. Nicholas Alonso and Jeff Krichmar from UC Irvine introduce a #neuralnetwork that implements an effective, principled approach to local, #online-continual #learning on #associative #memory tasks. #Brain-inspired #Computing and #Algorithms Now out 👉 #naturecommunications Nature Portfolio https://lnkd.in/etfqNu2z

62

1 Comment

I've been getting alot of push on metadata lately with requests for information all parts of a piece of data. Even something as simple as a photograph can have a ever increasing amount of metadata associated with it. It is now standard practice for at least the photographer is recorded on a photo. But the more metadata, the better. Except the increase in metadata is an increase in time to curate that metadata. Metadata which may never be used by anyone. Or maybe in 10 years, someone needs tree stand density at the exact location and they could get it from your photograph, with a few pieces of metadata. But how does that investment in time get recouped?

15

5 Comments

The Publications Office of the European Union developed model2owl for the eProcurement Ontology Project, but other industries and standardisation bodies have since recognised its potential. Its ability to automate complex processes and ensure consistency across multiple artifacts has made model2owl a valuable asset in various domains. 1. Adoption by SEMIC and other standardisation bodies SEMIC (Semantic Interoperability Community) is an initiative by the European Commission to promote semantic interoperability across the EU. ✅ SEMIC’s endorsement of model2owl SEMIC recognised the need for a tool to standardize building semantic data specifications across the EU. They saw that model2owl provided a perfect solution for ensuring consistency and coherence across different semantic standards. ✅ Harmonising standards across the EU One of SEMIC’s key goals is to harmonise semantic standards across member states. By adopting the UML conventions implemented in model2owl, SEMIC could streamline the standardisation process, ensuring that UML diagrams were drawn and interpreted consistently across different projects. 2. Expanding model2owl’s capabilities There is increasing demand to generate semantic artifacts like OWL and SHACL and technical artifacts such as XSD (XML Schema Definition) and JSON (JavaScript Object Notation) schemas. ✅ Meeting developers' needs While model2owl’s primary function is to generate semantic artifacts, developers are also growing in demand for the tool to create technical artifacts. This would allow developers to take these artifacts and immediately start working with them without manually creating XSD or JSON schemas. ✅ Bridging the gap between semantic and technical interoperability layers This requires careful coordination between semantic engineers and developers to ensure the generated technical artifacts are consistent with the underlying semantic models. As model2owl expands its capabilities and gains recognition from standardisation bodies like SEMIC, it will play an increasingly important role in shaping the future of semantic data standards across industries. More information about the project here 👇 https://lnkd.in/eJKYg6js Semantic engineers are more than welcome to help expand and mature this project. If you are interested in contributing, please DM me. I will be happy to share all the information. -- model2owl is the property of the Publications Office of the European Union, and Meaningfy was honored to have contributed to the development of this project by providing semantic expertise and technical implementation. Meaningfy continues to support the European Commission’s initiatives, leading the charge toward a transparent, efficient, and interconnected European public sector. If you represent a European Institution or a public company that needs to implement an interoperability solution, contact us, and together, we can find the best solution.

20

Bit by bit, we're slowly reinventing the semantic web. The article here explores an idea that I've been thinking about as well. There are two key advantages to integrating knowledge graphs via RAG vs. trying to work JUST from LLMs - the fist is that you have the ability to name entities and taxonomies, ensuring topical synchronization via taxonomies. However, a second aspect is that you can teach an LLM to utilize a specific data structure by passing it a schema. However, there's a corollary to that. If you can teach an LLM how to build schemas, then the LLM can use this to design a schema for its own purposes to optimize reasoning. Now, I'm going to make a contrafactual argument here to others (people I respect) to say that I think an LLM is perfectly capable of reasoning, because we don't understand reasoning. An LLM is incapable (and even here I'd quibble) to utilize intent, but if it is specifically informed to create a "castle in the sky" graph that allows it to perform reasoning tasks, it should be able to traverse that graph with appropriate constraints, because I've used that technique myself when using LLMs to design schemas. Put another way, in order for an LLM to reason, it needs to become an ontologist. https://lnkd.in/gW6D9TeW

45

6 Comments

Yet another indicator about why LLMs are not databases. They are non-deterministic (in other words, they are stochastic), meaning that even with the same prompts and contexts, different host GPUs can change the output of the LLM in different ways. Even with RAG, replicability is not guaranteed. The underlying message here? LLMs value as generators is unquestionable, and it is a good tool for natural language prompts. However, LLMs likely will never be reliable enough to serve as databases, precisely because they are chaos machines: they are extraordinarily dependent upon starting conditions. https://lnkd.in/gBDei_ye

277

35 Comments