Multi-Agent LLMs for Automatic Ontology Generation: A New Paradigm in Knowledge Engineering
A latest arXiv study proposes an automatic ontology generation method based on multi-agent large language models. Throug…
📂 Research
Latest articles in Research
Analytica: Enhancing LLM Analytical Capabilities with Soft Propositional Reasoning
A research team has proposed the Analytica agent architecture based on Soft Propositional Reasoning (SPR) principles, re…
Can Wi-Fi Signals Read Human Behavior? New Research Achieves Interpretable Activity Recognition
Researchers propose a novel framework combining discrete latent variable compression with Linear Temporal Logic rule ext…
New Research Proposes Systematic Debugging Methods for Large Language Models
A latest arXiv paper proposes a systematic debugging framework that treats large language models as 'observable systems,…
Decoupled Human-in-the-Loop Systems: Making AI Agents Safer and More Controllable
A latest arXiv paper proposes a decoupled Human-in-the-Loop (HITL) system architecture that separates human oversight me…
FormalScience: Using AI Agents to Automatically Formalize Scientific Reasoning into Lean-Verifiable Code
A research team has released the FormalScience framework, which leverages human-in-the-loop collaboration and AI agent c…
The Power of Power Law: Data Imbalance Actually Enhances AI Reasoning Capabilities
A new study finds that the power-law distribution inherent in natural language is not a barrier to model learning but ca…
PExA: Solving the Text-to-SQL Latency Challenge with Parallel Exploration
A latest arXiv paper proposes PExA (Parallel Exploration Agent), which draws on software test coverage concepts to decom…
Multi-Fidelity Digital Twin Empowers Intelligent Fault Diagnosis for General Aviation
A latest arXiv paper proposes an intelligent fault diagnosis framework for general aviation aircraft based on multi-fide…
Human-1: The First Full-Duplex Conversational AI System for Hindi
The Josh Talks team has released the Human-1 framework, adapted from the Moshi architecture for Hindi. Trained on 26,000…
New Training-Free LLM Context Compression Method: Hybrid Graph Priors
A latest arXiv paper proposes a training-free context compression framework based on hybrid graph priors. Moving from si…
S²IT: Stepwise Syntax Integration Tuning Enhances LLM Aspect Sentiment Quad Prediction
A latest arXiv paper proposes the S²IT method, which effectively incorporates syntactic structure information into the g…